This repository is a beginner-friendly TensorFlow 2 introduction that demonstrates how a neural network can learn a simple linear relationship from data.

The notebook uses the equation:

y = 2.5x + 3

to show the difference between traditional programming and machine learning, and then builds a tiny TensorFlow model to learn that relationship automatically.

• Project Overview
• Files in This Repository
• What the Notebook Covers
• Visual Insight
• Code Example
• Why This Project Matters
• How to Run
• Future Improvements
• Author

The goal of this project is to explain the core idea of machine learning with a very simple example.

Instead of hardcoding the equation directly, the notebook:

• prepares example input-output pairs
• creates a neural network with one dense layer
• compiles the model
• trains it over multiple epochs
• visualizes the training loss
• uses the trained model to make predictions

This makes the repo a nice introduction to:

• TensorFlow 2
• Keras Sequential models
• supervised learning
• loss minimization
• prediction with trained models

• MLWithTensorflow2.ipynb: main notebook
• images/tensorflow_learning_target.svg: visual created from the notebook’s learning target
• README.md: project documentation

The notebook is structured as a short learning walkthrough:

It starts by showing how the equation can be solved directly with standard code.

It then reframes the same problem as a learning task:

• give the model example values of x
• provide matching outputs y
• let the model learn the mapping

The notebook creates a simple dense layer:

• one input
• one output

This is the smallest useful neural network for learning a straight-line relationship.

The model is compiled with:

• loss: mean_squared_error
• optimizer: Adam(0.1)

and then trained for 100 epochs.

Finally, the notebook:

• plots training loss
• tests predictions on new values

This visual shows the exact input-output relationship used in the notebook. It helps make the project immediately understandable on GitHub before opening the notebook.

The key idea is simple:

• blue points represent sample training data
• the orange line represents the pattern the model is trying to learn

This is the central modeling block from the notebook:

import tensorflow as tf
import numpy as np

x = np.array([-10, -5, 0, 2, 100], dtype=float)
y = np.array([-22, -9.5, 3, 8, 253], dtype=float)

layer = tf.keras.layers.Dense(units=1, input_shape=[1])
model = tf.keras.Sequential([layer])

model.compile(
    loss='mean_squared_error',
    optimizer=tf.keras.optimizers.Adam(0.1)
)

train = model.fit(x, y, epochs=100, verbose=False)
prediction = model.predict([50])

This example shows the full beginner workflow:

• define training data
• create a layer
• build a model
• compile it
• fit it
• predict on a new value

This project is useful because it explains TensorFlow through a very approachable example rather than a large complex dataset.

It demonstrates:

• the difference between explicit programming and machine learning
• how TensorFlow models are built with Keras
• how a model learns from examples
• how loss and optimization fit into training
• how predictions are made after fitting

It works especially well as a portfolio project because it shows that you can teach and implement machine learning concepts clearly.

1. Clone the repository.
2. Open the project folder.
3. Install the required libraries:

pip install tensorflow numpy matplotlib jupyter

4. Launch Jupyter Notebook:

jupyter notebook

5. Open MLWithTensorflow2.ipynb and run the cells in order.

You can also open it directly in Colab: Open in Colab

This project could be improved further by:

• exporting the notebook’s training-loss plot as a repo image
• adding a second example with a non-linear problem
• introducing train/test split concepts
• adding comments for each TensorFlow step in the notebook
• including a short “TensorFlow basics” section for first-time learners

Divya Thakur

• GitHub: DivyaThakur24
• LinkedIn: divya-thakurr
• Portfolio: divyathakur24.github.io/DivyaThakurPortfolio