Slow going in air travel….

The Task

The aim for this project is to: - Uncover the operational factors which are most significant in airport delays. - Identify airports that are most impacted by delays. - Make recommendations regarding how to decrease delays.

The Methods

• Explore and analyze the ASPM data provided
• Utilize Principle Component Analysis (PCA) to distill several features to fewer
• Perform unsupervised cluster analysis to seek out patterns in the data.

The Process

I first needed a better understanding of the data from an operational point of view. I researched the definitions of the data provided based on the information available at the ASPM website. Next I decided on which features were most likely to provide information regarding the contributing factors of delays. I felt comfortable in using the following 5 features because they were all specifically measures of delay and not a straight time measurement: - average_gate_departure_delay
- average_taxi_out_delay
- average_airborne_delay
- average_taxi_in_delay
- average_gate_arrival_delay

The PCA

In order to reduce the complexity of the features, based on the correlation between some of my key features (see heatmaps below), I decided that a Principle Component Analysis would be a reasonable way to proceed.

I performed the PCA using a component value of 3. The results of the PCA demonstrated that, indeed, 94% of my variance could be explained by 3 components.

---