Lead scoring is an effective lead prioritization method used to rank prospects based on the likelihood of converting them to customers. While there are many available data that company gather, it is difficult to identify which attributes are important in determining a successful sale. This study proposes a method that can leverage the historical data to determine the likelihood of successful sales conversion and identify the important factors that contribute to this conversion.

The steps involved for this case study are mentioned below:

1. Data Loading
2. Data Cleaning & Pre-Processing
3. Split Training and Test Data (Holdout Method)
4. Run Logistic Regression for Full Model and Null Model
5. Run Forward and Backward Stepwise Regression based on AIC Selection Criteria
6. Evaluate Models through Accuracy, Specificity, Sensitivity & AUC.
7. Determine Optimal Threshold (Youden Index).

One can obtained a valuable business insight by filtering out large number of attributes and selecting few key important indicators to enhance the predictive lead scoring system. Companies can then focus their resources and marketing efforts to the right leads which ultimately improve the sales process