SAP has integrated machine-learning algorithms in SAP Analytics cloud as Predictive scenarios. It offers a simple user-friendly interface to perform predictive analytics
In these series of blogs, I would like to discuss currently available models Regression, Classification, Time series.
Regression Analysis: This is a supervised statistical method for estimating the relationship between dependent and independent variables. It helps in predicting and forecasting dependent values.
For demonstration considered SAP AR training Datasource: SAP__FI_AR_IM_PAYMENTFORECASTINGHIST and in this context, I aim to find out what the factors influencing and impacting due dates of receivables.
- Select a Create predictive model from Menu option Predictive scenarios, and click on create predictive model option to create a new model. ( in this example I created mode with name Test_Regression)
2. Select dependent fields – As we are trying to predict overdue days select OVERDUEDAYS as Predictive Goal.
3. Select Independent fields – In our example, I exclude few influencers like Key data, document number as they are not key influential parameters on due dates.
4. Select necessary fields and click on train data model.
5. To know how well the model built, check regression output parameters and its influencers.
Root mean square error of 12.14 and predictive confidence of 99.75, which show results of the model on unseen data.
Influencer contributors show the % of correlation the independent value having on dependent values.
6. Apply data model to a new data set, which is SAP provided data set SAP__FI_AR_IM_PAYMENTFORECASTINGCURR. Create a new output file to apply the model on a new set of data and predict Overdue days.
Conclusion: we discussed how to perform regression analysis using SAC and usage. Although SAC provides a user-friendly way of leveraging ML models, it is not a full-fledged ML solution.