WEBVTT

00:00.040 --> 00:00.920
Welcome back.

00:00.920 --> 00:03.840
Did you see how important to visualize data like.

00:03.880 --> 00:05.800
Actual and predicted data.

00:05.840 --> 00:08.240
Also how to make this graph.

00:08.480 --> 00:11.840
So Matplot library is a very important library.

00:11.840 --> 00:21.000
I advise you to use it in many, uh, in many fields and in many projects in order to understand well

00:21.040 --> 00:26.960
your data, understand your model, understand your results and so on.

00:27.000 --> 00:37.520
Another metrics and results we can observe and display and visualize using matplotlib is the plot,

00:37.560 --> 00:40.880
training histories and the errors.

00:40.880 --> 00:45.720
So here we are plotting errors and loss.

00:45.720 --> 00:49.680
And a loss means the MSE.

00:49.840 --> 00:51.800
Let's start with plot.

00:51.920 --> 01:01.510
So whenever you want to create a new graph plot dot figure figure size ten by four plot subplot 1 to

01:01.510 --> 01:01.960
1.

01:02.000 --> 01:03.440
Plot history.

01:03.480 --> 01:04.200
Loss.

01:04.360 --> 01:04.960
Title.

01:04.960 --> 01:05.960
Model loss.

01:06.070 --> 01:10.070
label, last MC and label epic.

01:10.110 --> 01:16.190
Everything we've seen in the previous videos except the plot dot subplot.

01:16.230 --> 01:27.270
This creates a one row, two column grid of subplots and selects the first subplot, which is the left

01:27.270 --> 01:27.870
side.

01:27.870 --> 01:29.670
Let me run and here we go.

01:29.870 --> 01:36.150
We have the epic from 0 to 1000 and from 0 to 25.

01:36.190 --> 01:37.390
The loss in MSE.

01:37.710 --> 01:46.510
And you notice that the loss in MSE decreased sharply to roughly around zero and continue this straight

01:46.510 --> 01:46.750
line.

01:46.750 --> 01:48.510
So this is the last MSE.

01:48.550 --> 01:54.270
So as as you train your model as the result becomes better.

01:54.310 --> 01:57.830
Let me display the Mae.

01:57.990 --> 01:59.910
So plot subplot.

01:59.950 --> 02:04.070
We have to the right here and the same as before.

02:04.110 --> 02:12.050
The subplot creates one row two column and to select the right side and plot.

02:12.210 --> 02:12.970
History.

02:13.170 --> 02:15.570
History and model.

02:15.610 --> 02:17.650
May and X label.

02:17.690 --> 02:21.410
The epic Y label is May run.

02:21.410 --> 02:22.650
And here we go.

02:22.850 --> 02:25.890
We have the model loss and the model.

02:25.930 --> 02:38.650
May you see guys, the model may starts from five degrees to roughly around 0.5 and then decreased around

02:38.650 --> 02:40.050
to 0.2.

02:40.250 --> 02:40.730
Okay.

02:40.930 --> 02:44.330
You see guys the model and the model loss.

02:44.370 --> 02:47.050
Every epoch gets better.

02:47.050 --> 02:53.010
So 1000 epoch 1000 epochs gets better and better.

02:53.090 --> 03:02.090
Um results better trained, model loss decreased and the errors decreased.

03:02.290 --> 03:02.690
Okay.

03:02.730 --> 03:12.210
So this is how we analyze the model loss MSE and Mae and plot them onto those graphs.