Summary:

In this project, using a time series of Daily Minimum Temperatures in Melbourne from 1981 to 1990, a neural network model contaiining RNN and Conv1D layers was built to predict the Daily Minimum Tempereature for Melbourne in the future.
Here, TensorFlow was used for building the neural network model, and keras-tuner was used to automate the hyperparameter tuning process.
Mean Absolute Error (MAE) was used as the loss metric, and the model was able to achieve 2 MAE degree Celsius of less than on the validation data.