Data Science Project: Engineer Time Series Data For Classification With Machine Learning

In this data science project in Python, I walk you through my reasoning when I have to build a binary classification model, but the data is time series data or transaction data. From problem framing to feature engineering, data modeling and cross validation, I go step by step in this beginner friendly data science tutorial. Moreover, I explain how to split your dataset in train and test while insuring you don't have data leakage. I explain how I think about measuring the model performance and metric interpretation. Will a customer buy at least once in the next three days? 00:00 - Introduction 00:58 - Load a csv with pandas 01:57 - Problem framing 03:23 - Discard time information from datetime column 04:51 - Aggregate pandas data frame by two conditions 06:50 - Fill in the missing days in time series 13:38 - How to use time series data in machine learning 16:08 - Feature engineering from time series data 21:09 - What is data leakage and how to avoid it? 22:45 - Remove duplicate rows from dataframe by subset of columns 23:25 - Data balancing 26:02 - Split data in train and test 27:15 - Binary classification with XGBoost 29:10 - Metrics for binary classification 37:47 - Cross validation 42:20 - Improvement potential 43:02 - See you next time! #datascience #datascienceproject #machinelearning #machinelearningproject #timeseriesanalysis Dataset: https://data.mendeley.com/datasets/9j... Same dataset is used in: Part 1: Exploratory Data Analysis Tutorial:    • Exploratory Data Analysis In Python: Machi...   Give a 🌟 to the code repository: https://github.com/giraffa-analytics/...