Predicting customer spend with Python скачать в хорошем качестве

Predicting customer spend with Python

A tutorial on how you can use Python and AI to predict how much a customer will spend in your store the next time they visit. For more info on data science consulting services visit https://www.fastdatascience.com The dataset I used is from Kaggle, https://www.kaggle.com/c/customer-spe... You can see the notebook source code at https://github.com/woodthom2/predict_... I am using Random Forest Regressor in Scikit-Learn with Python and Jupyter Notebook, and my customer data is in a MySQL SQL database. If you wanted to make a very big production grade system the principle would be the same but you might use Spark or BigQuery for the customer data instead of MySQL, and you might use Spark MLLib or Azure ML for the machine learning and set it up to run as a batch job and produce predictions daily. You might also train a new model for unknown customers, where you take demographic information such as the average purchase for people in that income bracket, or postcode, where you have no history for that individual. ADAPTING THE MODEL FOR MORE SOPHISTICATED USE CASES This model predicts the spend of a customer given the fact that they have made a purchase. Depending on your business needs, you might want a different model which instead predicts the total monthly spend of every customer whether or not they enter the store - which could be zero. If you are wondering how this model could cope with anomalies such as the coronavirus pandemic, you could add a feature indicating the purchase volume yesterday by all customers in the region. This would react to large upticks or downticks across all customers. You can also train a time series model representing the state of the economy as a whole and use the output of that model as an input into this model.