Logging the training progress in a CSV

Let's see how we can log the progress and various metrics during the training process to a csv file. Let's first import some libraries

import keras
import numpy as np

In this example, we will be using the fashion MNIST dataset to do some basic computer vision, where we will train a Keras neural network to classify items of clothing.

In order to import the data we will be using the built in function in Keras :

keras.datasets.fashion_mnist.load_data()

The model is a very simple neural network consisting in 2 fully connected layers. The model loss function is chosen in order to have a multiclass classifier : "sparse_categorical_crossentropy"

Let's define a simple feedforward network.

##get and preprocess the data
fashion_mnist = keras.datasets.fashion_mnist
(train_images, train_labels), (test_images, test_labels) = fashion_mnist.load_data()
train_images = train_images / 255.0
test_images = test_images / 255.0

## define the model 

model = keras.Sequential([
    keras.layers.Flatten(input_shape=(28,28)),
    keras.layers.Dense(128, activation="relu"),
    keras.layers.Dense(10, activation="softmax")
])

model.compile(optimizer="adam",
             loss = "sparse_categorical_crossentropy",
             metrics = ["accuracy",'mae'])

In order to stream to a csv file the epoch results and metrics, we define a CSV logger. It is a callback located in

keras.callbacks

Let's first import it

from keras.callbacks import CSVLogger

We now need to define the callback by specifiying a file to be written to, the separator and whether to append to the file or erase it every time.

The callback has to be added to the callbacks list in the fit method.

csv_logger = CSVLogger(filename="my_csv.csv", separator=';', append=False)
model.fit(train_images, train_labels, epochs=5, callbacks=[csv_logger])

Epoch 1/5
60000/60000 [==============================] - 9s 148us/step - loss: 0.5020 - acc: 0.8234 - mean_absolute_error: 4.4200
Epoch 2/5
60000/60000 [==============================] - 8s 138us/step - loss: 0.3765 - acc: 0.8630 - mean_absolute_error: 4.4200
Epoch 3/5
60000/60000 [==============================] - 8s 129us/step - loss: 0.3371 - acc: 0.8789 - mean_absolute_error: 4.4200
Epoch 4/5
60000/60000 [==============================] - 8s 133us/step - loss: 0.3129 - acc: 0.8843 - mean_absolute_error: 4.4200
Epoch 5/5
60000/60000 [==============================] - 9s 151us/step - loss: 0.2952 - acc: 0.8916 - mean_absolute_error: 4.4200





<keras.callbacks.History at 0x1582adc6780>

The results are stored in the my_csv.csv file and contain the epoch results

import pandas as pd
pd.read_csv("my_csv.csv", sep=";")

	epoch	acc	loss	mean_absolute_error
0	0	0.823400	0.502013	4.42
1	1	0.863050	0.376516	4.42
2	2	0.878867	0.337097	4.42
3	3	0.884317	0.312893	4.42
4	4	0.891583	0.295157	4.42