p3: make visualise.py useful

p3/visualise.py

@@ -1 +1,66 @@
-print("visualise")
+import pandas as pd
+import numpy as np
+import seaborn as sb
+import scipy
+import matplotlib.pyplot as plt
+
+uy = pd.read_csv('data/m.csv')
+print(uy)
+u = uy['u']
+u = u[1:len(u)]
+y = uy['y']
+y = y[1:len(y)]
+
+
+plt.plot(u)
+plt.xlabel("$ \mathit{k}T $")
+plt.title("Signal u")
+plt.savefig('res/signal_u.png', dpi=300)
+plt.clf()
+# ----
+plt.plot(y)
+plt.xlabel("$ \mathit{k}T $")
+plt.title("Signal y")
+plt.savefig('res/signal_y.png', dpi=300)
+plt.clf()
+
+correlation=scipy.stats.pearsonr(uy['u'], uy['y'])
+print(correlation)
+
+p = sb.lmplot(
+		data=uy,
+		x='u',
+		y='y',
+		fit_reg=True,
+		order=2,
+	)
+p.set(title='Correlation UY')
+p.savefig('res/uy_correlation.png', dpi=300)
+plt.clf()
+plt.close()
+
+
+theta = pd.read_csv('data/theta.csv', header=None)
+#  theta = np.transpose(theta) # -> no need to transpose.
+
+th_columns = ["$ \hat{a}_1 $","$ \hat{a}_2 $","$ \hat{b}_1 $","$ \hat{b}_2 $"]
+for i in theta:
+	plt.plot(theta[i], label=th_columns[i])
+	plt.locator_params(axis='x', nbins=12)
+
+plt.legend()
+plt.xlabel("$ \mathit{k}T $")
+plt.title("ARX Theta (Θ) Parameter Estimation using Recursive Least Squares")
+plt.savefig('res/theta.png', dpi=300)
+plt.clf()
+
+
+error = pd.read_csv('data/estimate_error.csv', header=None)
+
+plt.plot(error, label="$ ê(\mathit{k}T) $")
+plt.locator_params(axis='x', nbins=12)
+plt.legend()
+plt.xlabel("$ \mathit{k}T $")
+plt.title("ARX RLSq Theta (Θ) Parameter Estimation Error")
+plt.savefig('res/error.png', dpi=300)
+plt.clf()