Parking Forecasting System & Dataset

About 2 years of parking lot usage data were collected on ThingSpeak using a system made up of an IR sensor and an ESP32 board. The dataset contains slot availability information with respect to timestamps. The dataset can be used for time-series prediction tasks. This dataset is available on kaggle and huggingface.

Two ML models were trained on the collected dataset and were used to predict the probability of slot availability individually. When compared, the Multi-Layer Perceptron predicts slots with more probability of availability compared to the Random Forest Classifier. Implementation is available at github.