5. Deep Learning for Sequences

Activity 5.01: Using a Plain RNN Model to Predict IBM Stock Prices

Solution

1. Import the necessary libraries, load the .csv file, reverse the index, and plot the time series (the Close column) for visual inspection:

   import pandas as pd, numpy as np
   import matplotlib.pyplot as plt
   inp0 = pd.read_csv("IBM.csv")
   inp0 = inp0.sort_index(ascending=False)
   inp0.plot("Date", "Close")
   plt.show()

   The output will be as follows, with the closing price plotted on the Y-axis:

   Figure 5.40: The trend for IBM stock prices

2. Extract the values for Close from the DataFrame as a numpy array and plot them using matplotlib:

   ts_data = inp0.Close.values.reshape(-1,1)
   plt.figure(figsize=[14,5])
   plt.plot(ts_data)
   plt.show()

   The resulting trend is as follows, with the index plotted on the X-axis:

   Figure 5.41: The stock price data visualized

3. Assign the final 25% data as test data and the first 75% as train data:

   train_recs = int(len(ts_data...