Sep 6, 2018 · 1 min read
Good job.
Few pointers:
Instead of using pandas see if you can directly use numpy.loadtxt(…, delimiter=’ , ’) for loading the CSV (since it has only numerical values)
It’d be great if you perform an EDA (Exploratory Data Analysis) on your data before any preprocessing (missing value treatment, binning and other standard treatments); standard industry practice.
Keras has an inbuilt one-hot encoding method : to_categorical(…)
Also alternative way to one-hot encode would be:
one_hot = [0] * len(label_list)
one_hot[label_list.index(label)] = 1
Let me know your thoughts on this… again. Great post.
