In this short series of guides, we'll be taking a look at a hands-on house price prediction. We'll be using Keras, the deep learning API built on top of TensorFlow to train a neural network to predict the prices of houses from the Ames Housing Dataset, based on the 79 features it provides for each house.
We'll start out the project like all Machine Learning projects should start out with - with Exploratory Data Analysis, followed by Data Preprocessing and finally Building a Deep Learning Model to fit the data we've explored and cleaned previously.
Perform Exploratory Data Analysis (EDA) on the dataset to understand your data. We'll be posing hypotheses and exploring the correlations between features and our target variable - the "SalePrice".
Preprocess data to make it ready for training models on. We'll be dealing with missing values, one-hot-encoding categorical variables, clean up data types, and split the data into a training and testing set, as well as perform feature scaling.
Build a Deep Learning Neural Network with Python and Keras, train it and evaluate it on the data we've been exploring and cleaning in the previous two sections.