Current financial technology is one of the hottest application of machine learning business scenarios, strong wind control algorithm engineer needs both roots "wind control business," external work, but also minor algorithm engineer market "machine learning" Strength scarce.
Mutual benefit and risk control modeling tutorial online more foreign stale data sets are used, the actual current domestic credit has long been out of business, meaning less modeling. Accordingly, the present Chat not used LendingClub, GermanCreditRisk other foreign credit data set. Chat using this data source Renmou Ping real credit data for a year when typhoon control director is himself, complete code has been teaching on-line and stable operation code.
To ensure data compliance and legality of the use, it has been removed from sensitive user information such as the four elements, retaining only the public part of the network data and business metrics reptiles party data, etc., and the key variables have to do desensitization treatment. To ensure that the code does not involve the disclosure of trade secrets, for example, show only the core function of coding bins, WOE, IV, etc., does not affect the overall understanding of modeling. The above data, index, code, only for the current scorecard teaching, it is noted.
The main content of this field Chat 6:00 the following:
Consumer finance business scorecard, which is commonly used data source What? Scorecard model which indicators usually use? What are the characteristics of the project? How do the characteristics of financial engineering application scorecard?
How scorecard variables binning? Why variables that need to be binned? Commonly used methods such as binning: bin frequency, etc., and other wide-binning, chi-square Merge box, best decision tree based on the optimal binning, binning concept of what they are? There were among them how the advantages and disadvantages?
3. What is WOE and IV? What is the nature of business and WOE IV behind that? WOE calculated and IV is like?
4, behind the score card contains the kind of mathematical meaning? How LR by deriving a formula to quickly understand the essence of the scorecard?
5, how to convert between the model and the integrated scorecard model results? How deduced by mathematical formulas, the integrated model into fractional output?
6, how to evaluate the results of the model? How to use the "vernacular" to understand the meaning of KS and AUC values?
Read more: http://gitbook.cn/gitchat/activity/5cfe6073e096977d1d957281
A field watching too much trouble? GitChat become members and enjoy 1000+ games Chat! Click to view