Data is key
When it comes to improving the performance of a neural network, or any other machine learning model for that matter, the importance of good data preparation cannot be overemphasized. In Chapter 3, Linear Regression with TensorFlow, we saw the impact that normalizing our data had on the model’s performance. Beyond data normalization, there are other data preparation techniques that can make a difference in our modeling process.
As you must have recognized by now, machine learning requires investigating, experimenting, and applying different techniques, depending on the problem at hand. To ensure we have an optimally performing model, our journey should start by looking at our data thoroughly. Do we have enough representative samples from each of the target classes? Is our data balanced? Have we ensured the absence of incorrect labels? Do we have the right type of data? How are we dealing with missing data? These are some of the questions we have to ask and handle...