Infographic: Steps to Building a Statistical Model

Steps to Building a StatisticaL Model LEVEL 1 Business Problem • Abstraction of a business problem to a data science problem 1 LEVEL 2 Data ExPloration Exploration of fields in the data set and understanding what they Data collection from reliable sources mean checking the data for any abnormalities like missing values Making sese of the data using graphics and descriptive statistics LEVEL 3 Data Cleaning Imputation of missing values Fixing outliers LEVEL 4 Selection of the model Deciding on the type of model or models to be used based on the nature of problem and data 中 • calculating derived variables uhere required making sure that the model and variables adequately capture the business problen 0000 LEVEL S Training, testand validation sets Training the model using the "training" set Dividing data into Training, Test and Validation sets Checking goodness of fit criteria on both the training set & the test set. Repeating steps 2 and 3, by tuning model parameters & creating new variables if required, till good resuLts are obtained on both the sets checking if the model fits the data by applying the model on the "test" set 臭 吳 吳 Validation on the out-of-sample "validation" set Visit us at: AUREUS UUW.aureusanalytics.com ANALYTICS WWw.facebook.com/aureusanalytics www.aureusanalytics.com WWw.linkedin.com/company/aureus-analytics Disclainer: The nateriat has been reproduced and coRnunicated you. The naterial nay be subject to copyright under the Lau. •000 00000 00000