WEBVTT

00:00.080 --> 00:00.920
Welcome back.

00:00.920 --> 00:03.880
In this video we're going to talk about TensorFlow implementation.

00:03.880 --> 00:09.720
I want from you to focus with me because those are the blueprint for our application.

00:09.960 --> 00:15.640
Our linear relationship is price equals to three times size plus two plus noise.

00:15.680 --> 00:18.320
Here we have the size as an input.

00:18.480 --> 00:21.720
For example 1500ft².

00:21.800 --> 00:23.480
We have the input layer.

00:23.640 --> 00:26.560
The input layer takes only one number.

00:26.560 --> 00:29.240
So this is a very simple model.

00:29.240 --> 00:35.360
Other models takes many many numbers and many inputs like number of rooms like that.

00:35.360 --> 00:36.440
The position.

00:36.440 --> 00:42.960
But here in this example we are taking only the size in square feet as an example.

00:42.960 --> 00:52.040
Then we pass it in the for the other layer, which is containing only one neuron dense with one neuron.

00:52.160 --> 01:01.320
And here we have the equation y equal to w times x plus b slope times x variable plus the noise and

01:01.320 --> 01:03.120
or the intercept.

01:03.160 --> 01:03.640
Okay.

01:03.920 --> 01:08.120
Then after this layer we have the output layer.

01:08.120 --> 01:10.080
We need to get the price.

01:10.080 --> 01:18.820
So this price will lead after the that calculations and the machine learning and algorithms, we get

01:18.820 --> 01:22.020
the price $6,500.

01:22.060 --> 01:26.740
Okay, so this will create a linear stack of layers.

01:26.780 --> 01:32.980
Layers are executed in order from input to output, like stacking Lego blocks.

01:33.100 --> 01:42.980
And creating those layers is done by the Keras dot sequential and that TensorFlow Keras dot sequential

01:42.980 --> 01:43.660
function.

01:43.660 --> 01:46.060
Also to make the layer dense.

01:46.300 --> 01:47.420
Dense layer.

01:47.420 --> 01:48.620
Fully connected layer.

01:48.620 --> 01:51.580
Every input connects to every output.

01:51.580 --> 01:54.900
So one number of neurons in the layer.

01:54.900 --> 02:03.380
Here we have only one neuron okay one number of neurons in the layer which is the output dimension.

02:03.580 --> 02:06.340
We have to get the price.

02:06.340 --> 02:11.220
For linear regression one neuron equals to one output value.

02:11.220 --> 02:20.260
So here since we are dealing with linear relationship linear regression one neuron equals to one output

02:20.260 --> 02:22.820
value the predicted price.

02:22.820 --> 02:28.020
So here we are expecting one feature per sample.