Taught by: Snehan Kekre, Machine Learning Instructor, Machine Learning
In this project, I learned to use Azure Machine Learning Studio to build a predictive model without writing a single line of code. I predict flight delays using weather data provided by the US Bureau of Transportation Statistics and the National Oceanic and Atmospheric Association (NOAA).
- Build a predictive model using Azure ML Studio.
- Demonstrate a working knowledge of setting up experiments on Azure ML Studio.
- Operationalise machine learning work flows with Azure's drag-and-drop modules.
- Importing the Data Sets
- Scrubbing Missing Values
- Eliminating Target Leaks
- Conversion to Categorical Features
- Preparing Features to be Joined with Weather Data
- Preprocessing the Weather Dataset
- Joining Both Datasets
- Training and Evaluating the Model
Accuracy: 76.9%
- Figure: ROC Curve
- Summary of the model
- Applied Two-class logistic Regression to predict the model
- Able to train and evaluate a predictive model on Azure Machine Learning Studio, all without writing a single line of code!
- Able to predict flight delays using weather data provided by the US Bureau of Transportation Statistics and the National Oceanic and Atmospheric Association (NOAA).