KR101975448B1 - Evaluation System and Method for Big Data Based Commodity Investment Recommendation Algorithms Using Artificial Intelligence - Google Patents

Abstract

The present invention relates to a system which establishes big data with regard to products of common series such as real estate, stock products, funds, bonds, lotto, sports Toto and the like and estimates a hit rate by using artificial intelligence for an algorithm which recommends a product having a high investment value among the plurality of products. The system comprises: an element evaluation unit calculating an evaluation value of element data for each product; an expected ranking calculation unit calculating a sum score by summing each of the element data after applying weight to each of the element data for each product and setting an estimated profit ranking of the product in order of the summed score; and a hit rate calculation unit calculating a hit rate of the estimated profit ranking by using an actual profit ranking and the estimated profit ranking for each product.

Description

TECHNICAL FIELD [0001] The present invention relates to a system and a method for evaluating a recommendation algorithm based on Big Data based on Artificial Intelligence

The present invention relates to a system and a method for evaluating a big data-based commodity investment recommendation algorithm using artificial intelligence, and more particularly, to a system and a method for constructing big data for commodities of common series such as real estate, And a system for recommending a commodity having a high investment value among a plurality of commodities using an artificial intelligence driven according to a hit ratio algorithm.

A variety of analytical techniques have been proposed to select promising future products, such as real estate, stocks, funds, bonds, lotto, and sports Toto, that can be predicted and invested.

For example, a parametric model such as a linear regression model was used to estimate or determine the value of conventional real estate. However, there is a problem that the reliability of the estimated price is limited because there are many assumptions such as the regularity of the data, the independence of the explanatory variables, and the linearity of the price function. Real estate prices have various nonlinear factors that can not be determined only by the parameter model, but they can not be reflected in the existing models. For example, the trading unit price tends to decrease as the land area increases, but in the parameter model, this feature can not be reflected due to the linearity assumption. In addition, the price of real estate varies widely depending on its physical location, which is called spatial dependence. In the past, the characteristics of real estate price formation, that is, the factors of physical location and distance, were not reflected in the estimation of real estate prices.

Thus, a system for predicting the future value of a commodity such as a real estate, a stock commodity, a fund, a bond, a lottery, or a sports Toto requires an evaluation system for verifying the effect. However, There are not many.

Furthermore, it is obvious that the element data necessary for judging the future value of the product has an actual influence on the judgment of the future value and the magnitude of the influence, , Systems with these standards are in short supply.

Korean Patent Publication No. 10-1794027

SUMMARY OF THE INVENTION The present invention has been made in order to solve the problems of the related art as described above, and it is an object of the present invention to construct big data for commodities of common series such as real estate, stock, And a system for recommending a product having a high investment value among a plurality of products using artificial intelligence that is driven.

According to an aspect of the present invention, there is provided a system for evaluating a Big Data-based Product Investment Recommendation Algorithm using artificial intelligence according to the present invention, comprising: an element evaluation unit for calculating an evaluation value of element data for each product; An expected ranking calculation unit for calculating a sum score by summing up the respective element data after applying weights to the respective element data for each of the goods and setting an estimated profit ranking of the goods in order of the sum of the sum scores; And a hit rate calculating unit for calculating a hit rate of the estimated profit ranking using the actual profit rank and the estimated profit rank for each product.

The hit rate calculation unit may calculate a difference value by subtracting the real profit order for each product from the expected profit rank for each product, and calculate an actual sum value of the difference values of the goods.

The hit rate calculation unit may calculate the hit rate of the estimated profit ranking using the worst-case estimated sum value and the actual sum value calculated on the assumption that the difference value is greatest for each product.

Also, the worst expected value BEV is the number of the difference values (number of products) (n)

And calculating the value by substituting.

The apparatus may further include a weight computing unit for deriving a weight for each element data when the hit rate is the highest.

The weight calculator may set the number of the predetermined range for each element data as a weight, and search for the optimal weight for each element data using artificial intelligence.

The apparatus further includes an influence testing unit for replacing the weight applied to the selected element data with the largest or smallest value among the predetermined number of ranges and analyzing the influence of the selected element data using the degree of the changed hit ratio .

According to another aspect of the present invention, there is provided a method of evaluating a Big Data-based Product Investment Recommendation Algorithm using artificial intelligence, the method comprising the steps of: (a) ; (b) calculating an aggregate score by adding a weight to each element data for each of the products, and calculating an estimated profit order of the goods in the order of the aggregate score; (c) calculating a hit ratio of the expected profit rank by using the actual profit ranking and the estimated profit rank for each product.

The step (c) may further include deriving a difference value by subtracting the real profit order for each product from the estimated profit rank for each product; Calculating an actual sum value of the sum of the difference values of the goods; Calculating a worst case estimated value on the assumption that the difference value is the largest for each product; And calculating a hit rate of the expected profit rank using the actual sum value and the worst expected sum value.

Also, the worst expected value BEV is the number of the difference values (number of products) (n)

And calculating the value by substituting.

According to the system and method for evaluating the Big Data-based Product Investment Recommendation Algorithm using the artificial intelligence according to the present invention,

First, since the products that are expected to have a high profit among the plurality of products are displayed in order, a higher profit rate can be expected in the investment.

Second, by searching the optimal weight using the hit rate calculation algorithm, it is possible to derive the best estimated profit rank corresponding to the classification of goods or the difference of classification.

Third, when real estate appraisal of the present invention is applied, new reliability indexes such as demand energy and charter energy position value which can be used as element data are proposed and the influence of the new index on the hit ratio is maximized.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram of a system (investment recommendation system) for evaluating a big data-based commodity investment recommendation algorithm using artificial intelligence according to an embodiment of the present invention;
FIG. 2 is a diagram illustrating an example in which elementary item data, an estimated profit rank, an actual return rate, and a hit ratio are displayed in a system for evaluating a big data-based product investment recommendation algorithm using artificial intelligence according to an embodiment of the present invention.
FIG. 3 is a graph showing the demand energy of the Seoul area and the Gyeonggi area during the period from 1999 to 2020 in the embodiment utilizing the present invention for real estate evaluation.
FIG. 4 is a graph showing a trading index, a rent index, and charter energy in the Seoul area for the period from 1999 to 2018 in the embodiment where the present invention is applied to property evaluation.
FIG. 5 is a diagram showing an example in which element data for each real estate, an estimated profit rank, an actual return rate, and a hit ratio are displayed in an embodiment utilizing the present invention for real estate evaluation.
6 is a flowchart of a method for evaluating a Big Data-based Product Investment Recommendation Algorithm using artificial intelligence according to an embodiment of the present invention.
7 is a flowchart showing a detailed procedure of step S110.
8 is a flowchart showing a detailed procedure of step S114.
FIG. 9 is a graph showing the relationship between the real rate of return and the average average rate of return and the rate of rent of 70% or more after 2 years, when the investment is made in the top ranking (top 30) among the estimated profit rankings derived using the investment recommendation algorithm according to the embodiment of the present invention A graph comparing the yields of apartments.

A system and method for evaluating a Big Data-based Product Investment Recommendation Algorithm using artificial intelligence according to embodiments of the present invention will be described in detail with reference to the accompanying drawings. The present invention is capable of various modifications and various forms, and specific embodiments are illustrated in the drawings and described in detail in the text. It is to be understood, however, that the invention is not intended to be limited to the particular forms disclosed, but on the contrary, is intended to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the invention. Like reference numerals are used for like elements in describing each drawing.

Also, unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Terms such as those defined in commonly used dictionaries are to be interpreted as having a meaning consistent with the contextual meaning of the related art and are to be interpreted as either ideal or overly formal in the sense of the present application Do not.

The system and method for evaluating the Big Data-based Product Investment Recommendation Algorithm using artificial intelligence according to an embodiment of the present invention can be implemented by including a major component or function in any computer device or distributed computing system.

A system and method for evaluating a Big Data-based Product Investment Recommendation Algorithm using artificial intelligence according to an embodiment of the present invention is included in an investment recommendation system 100 for calculating an expected profit rank of a product, Evaluate the hit ratio.

1 and 2, a system for evaluating a Big Data-based Product Investment Recommendation Algorithm using artificial intelligence according to an embodiment of the present invention includes an element evaluation unit 110 for calculating an evaluation value of element data for each product, An anticipation ranking calculator 120 for calculating a sum total by adding weights to the evaluation values of the respective element data for each of the commodities and summing them and setting an anticipated profit ranking of the commodities in the order of the total sum score, And a hit rate calculation unit 130 for calculating a hit rate of the estimated profit rank using the actual profit rank of each product and the estimated profit rank.

Commodities include all types and intangible assets that can be subject to sale, such as real estate, stocks, funds, bonds, etc., or goods that need to be predicted, such as lotto or sports toto.

Element data constituting the name of the commodity and the characteristics of each commodity are inputted to the system through the input unit 160. [

The element data of the input goods is stored in the storage unit 180. The element data includes vast amounts of data such as sales prices accumulated over several decades for each product, so the element data can be regarded as Big data relating to the product.

In addition, information such as the estimated value of the elementary data for each product, the estimated profit rank, the actual profit rate, and the hit rate can be output using the display unit 170. The display unit 170 may be a known information output device such as a monitor or a printer.

The investment recommendation system 100 preferably calculates an expected revenue rank for one common affiliate product. A common series may be, for example, an apartment (real estate) located in a specific area, a stock commodity corresponding to the same business type, or a fund investing a similar commodity.

The hit ratio calculation unit 130 for calculating the hit rate of the expected revenue rank for each product can use the AI algorithm for rapid analysis of vast amounts of data.

Hereinafter, the present invention will be described using an embodiment in which the investment recommendation system 100 is applied to the field of real estate.

Real estate (commodity) is a category of real estate such as land, house (apartment), etc., and includes objects that can be traded.

Each property contains element data in detail. The factor data may include the equilibrium of apartment complexes, the rent rate, the investment gap (difference between the selling price and the rent price), the yearly (open or inaugurated year), the number of households, the 6 month price increase rate,

Element data can include demand energy. Demand energy is an indicator of the adequacy of the supply of apartment households in an area. The demand energy is a cumulative sum of the demand quantity and the supply quantity difference. If the supply quantity is large, the value of the demand energy decreases. When the supply quantity is small, the value of the demand energy increases. At this time, the demand amount is calculated as the product of the population number and the household life cycle. The value of the household life cycle may be 0.5%, which is normally calculated as the demand in a year. If you want to calculate demand by Will, you can divide this value by 12. The supply amount is calculated as the time when the real estate (housing) is completed in the area and can be moved.

3 (a) is a graph showing the demand energy in the Seoul area. This graph shows the demand energy flow, which is calculated as the cumulative sum of monthly demand and supply differentials since 1998. Demand energy has continued to increase due to shortage of supply from 2005 to 2016, but demand has remained in place due to a large increase in supply from 2016.

3 (b) is a graph showing demand energy in the Gyeonggi-do area. In Gyeonggi Province, the supply of energy has increased from the year 2005 to the end of 2016 due to supply shortage. However, after 2020, demand energy is rising again due to lack of additional supply. At this time, it can be seen that the demand energy of 2020 is about the level of May 2013.

The higher the demand energy, the lower the supply amount. Therefore, the higher the possibility of the increase of the selling price, and the lower the supply amount, the lower the possibility of the increase of the selling price.

Demand energy has a relatively higher impact on hit rate than real estate in the metropolitan area when evaluating real estate outside the Seoul metropolitan area. Therefore, demand energy can be regarded as relatively more important factor data when evaluating real estate outside the Seoul metropolitan area.

Demand energy can be applied not only to cities and provinces but also to sub-cities.

Demand energy can be used as an important measure as well as the current position value and the rate of increase (slope). Table 1 shows the calculated hit ratios after applying the weight to the current position value or rate of increase of the demand energy.

Time Hit rate of current position value (%) 200801 60.84 200901 62.12 201001 68.48 201101 58.21 201201 72.15 201301 60.67 201401 66.45 201501 51.72 201601 59.26 Average 62.21

In addition, the element data may include a value of the energetic energy position. FIG. 4 (a) is a graph of a trading index and a lease index of Seoul, and FIG. 4 (b) is a graph of a lean energy position value obtained by cumulatively calculating a difference between a lease increase and a sales increase in Seoul.

The charter energy location value is a cumulative sum flow of rise in the price of a house and rise in a sale price. In addition, the charter energy current position means the level of the current position relative to the maximum value of the energetic energy position value. The location of the charter energy present can be expressed by the following equation.

[Equation 1]

는 현재 전세에너지,

는 최대 전세에너지,

는 최저 전세에너지이다.

Currently, charter energy,

The maximum charter energy,

Is the lowest charter energy.

Real demand is a demand for residential purchase or rental housing without consideration of future value increases. On the other hand, the purchase price is a type of purchase that invests for the purpose of raising future value beyond the actual residence purpose. The demand for rent, the demand for rent, and the demand for rent, which reside in the premises, appear as a tendency to tug each other.

It is the graph of the energetic energy to visually show the demand and the actual demand. Seoul is showing signs of a downward trend when charter energy is rising, leading to a surge as charter energy declines and discharges, and when too much charter energy falls. On the other hand, in cities outside the Seoul metropolitan area, which is mainly focused on actual demand, it is uncommon for charter energy to drop below zero.

The value of lease energy location has a relatively high impact on the hit rate in real estate evaluation in the metropolitan area than the real estate outside the metropolitan area. Therefore, the lean energy location value can be included in the element data only in the real estate evaluation in the metropolitan area.

Table 2 shows the calculated hit ratios after applying the largest weight to the energies.

Time Of the energetic energy location value
Hit ratio (%) 200801 71.64 200901 80.73 201001 84.53 201101 83.36 201201 68.49 201301 73.02 201401 70.11 201501 70.07 201601 79.41 Average 75.71

As the weighted energy of charter energy is higher, the rate of hit is higher than that of demanded energy. Therefore, the value of charter energy is more important than demand energy. Therefore, when deriving the expected profit rank, the value of the charter energy position is weighted more than the demand energy.

In addition, the factor data includes the household growth rate, the rental rate, unsold housing units, the unsold pre-construction units, the volume of transactions compared with the previous year, and the amount of nearby residential units after two years.

The growth rate of households is obtained by using the number of households in the districts provided by the National Statistical Office to find out how much the number of households in the area increased in the previous year.

The rent rate is the ratio of the current market price to the current market price. It is the ratio of the average selling price to the charter price for each apartment, and the figure is obtained by the average value of the unit and the average value of the area (city and city).

On the other hand, the neighborhood position of the rent rate shows the position of which one real estate occupies in the regional average rent rate. At this time, the area means eup, myeon, and dong. The formula for calculating the rental rate neighborhood location is as follows.

&Quot; (2) "

는 현재 부동산의 전세가율,

는 지역 최대 전세가율,

는 지역 최저 전세가율이다. 일반적으로 시·구 단위의 지역은 전세가율은 높을수록 유리하다.

The current rate of rent,

The largest local charter rate,

Is the lowest minimum lease rate in the region. Generally, the higher the rent rate, the more favorable the area in city and district.

Unsold apartments and unsold units after completion are data provided by the Ministry of Land, Transport and Maritime Affairs.

The volume of transactions compared to last year is the increase rate of transaction volume compared to the same month of last year.

After two years, the amount of the moving-in area means the amount of the moving-in area at the time of the expiration of the two-year period in the city and district area adjacent to the area and the surrounding area.

[Table 3] and [Table 4] show the results of calculating the hit rate by setting the weight to 100 for any one of the household growth rate, the rental rate, the unsold housing unit, the unsold housing unit after completion,

Time % Of household growth rate Percentage of trading volume (%) compared to last year Hit rate (%) of surrounding residential area after 2 years 200801 61.05 65 61.74 200901 54.24 54.6 59.25 201001 56.94 59.78 65.04 201101 60.62 49.02 59.45 201201 63.7 61.07 58.64 201301 67.91 49.59 52.96 201401 70.84 62.94 48.79 201501 56.78 52.15 57.1 201601 58.47 56.03 55.78 Average 61.17 56.69 57.64

Time Unsold
Hit ratio (%) After completion of unsold housing
Hit ratio (%) Charter rate
Hit ratio (%) 200801 57.28 57.76 69.36 200901 61.26 61.27 75.96 201001 66.3 64.31 88.26 201101 64.72 74.82 91.01 201201 55.75 66.35 81.77 201301 65.71 63.85 86.28 201401 67.69 70.25 75.62 201501 66.69 60.87 68.39 201601 64.99 65.38 72.82 Average 63.38 64.98 78.83

The rent rate is higher than other factor data. 2015 was also lower than other periods, but it was very high in 2010 and 2011.

The estimated ranking calculation unit 120 calculates the total score by summing up the evaluation values to which the weights are applied after applying the weights to the evaluation values of the respective element data for each real estate, .

[Table 5] In applying the weights included in the predetermined range to the evaluation values of the element data, the weight computing unit 140 obtains weight values that can be commonly applied to the element data of all the years of the specified product by element data The post hitting rate was calculated. It can be seen that a good hit rate result can be obtained by applying an optimal weight set to a designated product group.

Time Hit ratio (%) 200801 81.53 200901 85.07 201001 88.80 201101 88.61 201201 83.59 201301 86.65 201401 81.50 201501 82.87 201601 85.21 Average 84.87

The influence testing unit 150 numerically indicates how much influence each element data has by using the hit ratio algorithm.

Also, by calculating the number of weight cases of each element data using the weight calculator 140, an optimal weight set that can derive the highest hit ratio can be obtained. That is, one embodiment of the present invention can be used as a tool for predicting a region or a real estate that is likely to rise at a time when patterns and weights of past element data of a product are used.

A product includes a plurality of element data, and a weight is set for each element data. Therefore, the weight set means a weight group corresponding to the number of element data.

In detail, the factor data for judging the real estate that is likely to rise in the region includes the equilibrium of the apartment complex, the rent rate, the investment gap (the difference between the selling price and the rent price), the yearly (open or in- Includes one-year growth rate.

Referring to FIG. 5, the element evaluation unit 110 may evaluate element data for each real estate, calculate evaluation values, and then list the evaluation values in a table format.

The evaluation value is a value calculated by a predetermined calculation method for each element data.

The optimal weight set is a value that can derive the highest hit ratio, and the weight calculation unit 140 searches.

The anticipation ranking calculator 120 may calculate the sum score using a method similar to Equation (4).

&Quot; (4) "

The expected revenue rank is set in the order of the number of products and the sum of the scores in ascending order.

The actual rate of return is calculated using the price of the commodity at the time when the estimated ranking calculation unit 120 sets the estimated profit rank and the price of the commodity after a predetermined time. And may be arbitrarily selected from one day to ten years after a predetermined time, or at a time when a price of a future commodity can be calculated, or thereafter. The hit ratio calculation unit 130 sets the rank order of products in descending order of the actual profit rate. The actual return rate can be the rate of price change from the point of time when the past estimated profit rank is set up to the evaluation point. However, if there are variables such as apartment price, Data such as the sum of households whose prices have risen can also be applied to actual returns.

The hit ratio calculating unit 130 calculates the hit ratio by comparing the estimated profit rank with the actual profit rank. Specifically, the hit ratio calculation unit 130 calculates a difference value by subtracting an actual profit order for each product from the expected profit rank for each product, and calculating an actual sum value of the difference values of the products. The difference value is represented by an absolute value. [Table 6] and [Table 7] are examples of calculation of the difference value and the sum value when the number of products is 100. [Table 6] shows the cases where the actual profit rank and the expected profit rank are perfectly the same, and [Table 7] is the case where the actual profit rank and the expected profit rank are completely different.

Estimated revenue rank Actual revenue rank Difference value One One 0 2 2 0 3 3 0 ... ... 0 98 98 0 99 99 0 100 100 0 Sum value 0

Estimated revenue rank Actual revenue rank Difference value One 100 99 2 99 97 3 98 95 ... ... 98 3 95 99 2 97 100 One 99 Worst Estimated Total Value 5000

[Table 6], if the actual profit order and the expected profit order are perfectly the same, the sum value becomes zero.

[Table 7], when the difference between the actual profit order and the expected profit order is completely different, the sum value is 5000.

The difference value is expressed as an absolute value, and the larger the difference value, the different the actual profit order and the expected profit order.

The difference value and the maximum value of the sum value are determined by the number of goods.

The hit rate calculation unit 130 calculates the hit rate of the estimated profit ranking using the worst-case estimated sum value and the actual sum value calculated on the assumption that the difference value is the greatest for each product.

(BEV, Bad expected value) is calculated by substituting the number of difference values (number of goods) (n) into Equation (5).

&Quot; (5) "

은 실제 수익 순위, a는 상품의 요소데이터의 개수,

는 요소데이터별 평가값,

는 요소데이터별 가중치이다.The hit ratio (HR, Hit ratio) is calculated by substituting the worst-case expected summation value (BEV) and the actual summation value into [Equation (6)]. Here, n is the number of items,

Is the actual revenue rank, a is the number of element data of the product,

Is an evaluation value for each element data,

Is a weight for each element data.

&Quot; (6) "

[Table 8] shows that the number of products is 100, the product with the first rank in the estimated profit rank is the 100th in the actual return ranking, the second in the estimated profit ranking is the first The sum of the values in the case.

Estimated revenue rank Actual revenue rank Difference value One 100 99 2 One One 3 2 One ... ... ... 98 97 One 99 98 One 100 99 One Actual sum value 198

Substituting the first case into (6), the hit ratio is 96.04%.

[Table 9] shows that the number of commodities is 100, the commodities with the first and second rank of expected profit are the 100th and 99th in the actual profitability rankings, respectively, In the second virtual case.

Estimated revenue rank Actual revenue rank Difference value One 100 99 2 99 97 3 One 2 ... ... ... 98 96 2 99 97 2 100 98 2 Actual sum value 390

Substituting the second case into (6), the hit ratio is 92.2%.

Meanwhile, one embodiment of the present invention further includes a weight computing unit 140 for deriving a weight for each element data when the hit ratio is the highest (highest hit ratio).

Specifically, the weight computing unit 140 sets a value of a predetermined range for each element data as a weight. In particular, the sum of the weights included in the weight set is always equal to the reference value. For example, when the predetermined range is 0 to 100, the element data is 5, and the reference value is 100, the weight of element data 1 is 10, the weight of element data 2 is 20, the weight of element data 3 is 30, The weight of the element data 4 is 15, the weight of the element data 5 is 25, and the total sum of the five weights is always set equal to the reference value.

The reference data is input to the system, the weight calculation unit 140 sets a weight set on the reference data, and the prediction ranking calculation unit 120 calculates a weight set for the reference data, And the hit rate calculation unit 130 calculates the hit rate in comparison with the actual profit ranking.

The reference data may be past element data information of a product, element data information of a similar product, and the like. For example, the past element data information of the product may be the element data of 2000 to 2017 of the corresponding apartment in order to calculate the expected profit rank of 2018 of A apartment or Z apartment. The element data of similar goods can be element data of Company B, which is listed on the KOSDAQ and which is similar to the business type and size in order to calculate the expected profit ranking of Company A newly listed on KOSDAQ.

The greater the element data, the wider the predetermined range, the higher the highest hit rate, but also the exponential increase in the combination of weights in the weight set. That is, it is desirable that a meaningful optimal weight set search is performed automatically since it requires a considerable computation time. In order to shorten the time for searching for an optimum weight set, the weight computing unit 140 includes an arithmetic function based on artificial intelligence.

)은 레퍼런스 데이터가 존재하는 모든 연도 개수(r), 상품의 개수(n), 상품의 요소데이터의 개수(a), 기 설정된 범위 및 기준값을 만족하는 가중치 세트의 경우의 수(M), 레퍼런스 데이터의 요소데이터별 가중치(

)를 [수학식 7]에 대입하여 산출할 수 있다.Highest hit rate (

(N), the number of element data (a) of the product, the number of cases (M) in the case of the weight set satisfying the preset range and the reference value, the reference number Weight of element data by data (

) Can be substituted into [Equation (7)].

&Quot; (7) "

)이 산출될 때의 요소데이터별 가중치(

)들이 최적의 가중치 세트가 된다.Highest hit rate (

) Is weighted by element data (

) Are the optimal set of weights.

The element data is a classification of a product, that is, a real estate, a stock item, and a fund product. However, when the classification of a product is real estate, for example, And the type and influence of factor data differ depending on the subdivision difference between eup, myeon, and dong. Thus, the optimal set of weights differs according to the classification and subdivision of the product. The weight computing unit 140 searches for an optimum weight set according to each situation.

In an embodiment of the present invention, the weight applied to the selected element data is replaced with the largest or smallest value among the predetermined range of numbers, and the influence of the selected element data is analyzed using the degree of the changed hit ratio And an influence testing unit 150 for analyzing the influence of the user.

If the highest or lowest weight is arbitrarily given to the selected element data to examine the influence of the element data, the calculated hit ratio differs from the highest hit ratio. The influence testing unit 150 analyzes the difference degree, .

Next, a method for searching an optimal weight set of a Big Data-based Product Investment Recommendation algorithm according to an embodiment of the present invention will be described.

Referring to FIG. 6, in the method of recommending a big data-based commodity investment, the weight calculator 140 calculates the number of weight cases of each element data by using the reference data corresponding to the commodity, and calculates the optimal weight A step S120 of inputting element data for each product by the input unit 160, a step S140 of calculating an evaluation value of element data for each product by the element evaluating unit 110, , Calculating a summed score by applying an optimal weight to each element data for each item, adding the estimated weight to each item data, and setting an estimated profit rank of the item in order of the summed score (S160 )Wow; And the display unit 170 displays the expected profit rank for each product (S180).

The calculation formula for calculating the highest hit rate in step S110 is as shown in equation (7).

7, in operation S110, the input unit 160 receives reference data corresponding to a product (S110). The weight computing unit 140 sets a value within a predetermined range for each element data as a weight Calculating a hit rate of the expected profit rank by using the estimated profit rank calculated by using the element data included in the reference data and the actual profit rank for each product, (S114).

The weight computing unit 140 determines whether the hit rate calculated in step S114 is the highest hit rate (S116). If the hit ratio calculated in step S114 is the highest hit ratio, the weight calculating unit 140 determines the set of weights used for calculating the highest hit ratio as the optimal set of weights (S118). However, if the hit ratio calculated in step S114 is not the highest hit ratio (S116), the process proceeds to step S112, and the weight calculation unit 140 sets a different weight set.

The calculation formula for calculating the hit rate in step S114 is as shown in equation (6).

Referring to FIG. 8, in operation S114, the element evaluation unit 110 calculates an evaluation value of element data based on reference data for each product (S1141), and a prediction ranking calculation unit 120 calculates an evaluation value of each element data A step S1142 of calculating an aggregate score by summing up the aggregated scores by applying the weights to the aggregated products, and setting an expected profit ranking of the products in the descending order of the aggregated scores, (S1144 to S1148) of calculating the hit rate of the expected profit rank using the profit rank.

The step of calculating the hit rate of the estimated profit rank by using the hit rate calculating unit 130 using the actual profit rank and the estimated profit rank may be performed by subtracting the real profit rank of each product from the estimated profit rank, (S1144) calculating an actual summed value obtained by adding the difference values of the goods, calculating a worst-case estimated sum value on the assumption that the difference value is greatest for each product (S1146) And calculating a hit rate of the expected profit ranking using the actual sum value and the worst estimated sum value (S1148).

The worst-case estimated sum value calculated in step S1146 is calculated by substituting the number (n) of difference values into the equation (5).

[simulation]

The yield rate when the investment recommendation algorithm according to an embodiment of the present invention is applied to the real estate (apartment) field is calculated.

The information of the investment recommendation algorithm is compared with the average average return rate and the return rate of apartments over 70%.

Using the multiple component data and the optimal weight set, we derive the expected revenue rank for each apartment for a period of 10 years and calculate the highest hit rate by comparing the actual profit rank two years later. The results of the calculated maximum hit ratio are shown in [Table 10].

Time Highest hit rate (%) 200801 81.53 200901 85.07 201001 88.80 201101 88.61 201201 83.59 201301 86.65 201401 81.50 201501 82.87 201601 85.21 Average 84.87

FIG. 9 is a graph showing the relationship between the real rate of return and the average average rate of return after a two-year period in the case of investing in the top ranking (the top thirty rank) among the estimated profit rankings derived using an investment recommendation algorithm according to an embodiment of the present invention, This is a graph comparing the yield of apartments. As shown in the graph, it can be seen that the ROI can be greatly improved by selecting an apartment to purchase using the investment recommendation algorithm according to an embodiment of the present invention.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments. It is clear that the present invention can be suitably modified and applied in the same manner. Therefore, the above description does not limit the scope of the present invention, which is defined by the limitations of the following claims.

100: Investment recommendation system 110: Element evaluation unit
120: Estimation Rank Calculation Unit 130:
140: weight calculation unit 150: influence test unit
160: input unit 170: display unit
180:

Claims (10)

An element evaluation unit for calculating an evaluation value of element data for each product;
A prediction ranking calculation unit for calculating a sum score by applying weights to the evaluation values of the respective element data for each of the products and summing them and setting an estimated profit ranking of the products in the order of the sum of the scores;
A hit rate calculation unit for calculating a hit rate of the estimated profit ranking using the actual profit rank of each product and the estimated profit rank; And
And a weight calculation unit for deriving a weight set when the hit ratio is the highest,
The weight computing unit sets the number of predetermined ranges for each element data as a weight, searches for an optimum weight set from which the highest hit ratio is derived using artificial intelligence,
And an influence testing unit for replacing the weight applied to the selected element data with the largest or smallest value among the predetermined number of ranges and analyzing the influence of the selected element data using the degree of the changed hit ratio A system for evaluating a commodity investment recommendation algorithm. The method according to claim 1,
Wherein the hit rate calculation unit derives a difference value by subtracting the real profit order for each product from the expected profit rank for each product and calculates an actual sum value obtained by summing the difference values of the goods system. 3. The method of claim 2,
Wherein the hit rate calculation unit calculates a hit rate of the expected profit rank by using the worst case estimated value and the actual sum value calculated on the assumption that the difference value is the greatest for each product, . delete delete delete delete (a) calculating an evaluation value of element data for each element evaluation apposition item;
(b) calculating an aggregate score by applying a weight to an evaluation value of each element data for each of the products, by summing up the aggregate scores, and setting an estimated profit ranking of the goods in descending order of the aggregate score;
(c) calculating a hit rate of the estimated profit rank by using the actual profit rank and the estimated profit rank by the product;
The weight computing unit sets the number of predetermined ranges for each of the element data as a weight and searching for an optimum weight set from which the highest hit rate is derived using artificial intelligence; And
Replacing the weight applied to the element data selected by the influence test section with the largest value or the smallest value among the number of the predetermined ranges and analyzing the influence of the selected element data using the degree of the changed hit ratio Lt; RTI ID = 0.0 > a < / RTI > product investment recommendation algorithm. 9. The method of claim 8, wherein the step (c)
Deriving a difference value by subtracting the actual profit order for each product from the expected profit rank for each product;
Calculating an actual sum value of the sum of the difference values of the goods;
Calculating a worst case estimated value on the assumption that the difference value is the largest for each product;
And calculating a hit rate of the expected profit ranking using the actual sum value and the worst estimated sum value. delete

Images