PREDICTING MOBILE PHONE

PRICING
Machine Learning Internship

Faza Ulfath – 1DB21CI022

1. Introduction

With the rapid advancement of mobile technology, the price of a mobile phone is influenced by several
factors such as processor speed, RAM, battery life, camera quality, and connectivity features.
Consumers often find it challenging to compare specifications and determine a suitable phone within
their budget. By using machine learning techniques, we can analyze mobile phone specifications and
predict their price category, providing valuable insights to manufacturers, retailers, and consumers.

This project aims to develop a machine learning model that classifies mobile phones into different
price ranges based on their technical specifications. The study focuses on exploring relevant features,
preprocessing data, training predictive models, and evaluating their performance.

2. Problem Statement

The pricing of mobile phones varies significantly depending on their specifications and brand value.
However, determining an accurate price category for a new mobile phone based on its features is a
challenging task. Currently, no straightforward method exists to estimate the price category of a phone
solely based on its technical specifications.

The objective of this project is to build a predictive model that classifies mobile phones into four price
categories: low cost (budget-friendly phones with basic features), medium cost (mid-range phones
with moderate performance), high cost (flagship phones with premium features), and very high cost
(luxury phones with cutting-edge technology). By developing a machine learning model, we aim to
provide a systematic and efficient approach to price prediction, which can be beneficial for
manufacturers in setting competitive prices and for consumers in making informed purchasing
decisions.

3. Objectives

The key objectives of this project are:

1. Data exploration to understand the dataset, identify trends, and analyze the impact of
different features on mobile phone pricing

2. Data preprocessing, including handling missing values, normalizing numerical features, and
encoding categorical variables

3. Feature selection to identify the most important specifications that influence price
classification

4. Model development by training and testing machine learning models to predict price
categories

5. Performance evaluation to compare different models and select the best one based on
accuracy and efficiency

6. Optimization by fine-tuning hyperparameters to improve predictive capabilities

7. Visualization and insights to present results through visualizations for a better understanding
of price trends in mobile phones

4. Expected Outcomes

By the end of this project, we expect to achieve a trained machine learning model capable of classifying
mobile phones into appropriate price categories. The study will help in identifying key mobile phone
features that significantly impact pricing. A comparative analysis of different machine learning models
will be conducted to determine the best-performing approach. The findings will be presented with
intuitive visualizations that highlight the relationship between mobile phone specifications and pricing
trends.

5. Dataset Description

The dataset used for this study consists of various mobile phone specifications and their corresponding
price categories. Each record in the dataset represents a different mobile phone model, with technical
features influencing its price classification.

Feature Description:

Feature Description

battery_power Battery capacity in mAh

blue Has Bluetooth or not (1 = Yes, 0 = No)

clock_speed Processor speed (GHz)

dual_sim Has dual SIM support or not (1 = Yes, 0 = No)

fc Front camera megapixels

four_g Has 4G or not (1 = Yes, 0 = No)

int_memory Internal memory in GB

m_deep Mobile depth in cm

mobile_wt Weight in grams

n_cores Number of processor cores

pc Primary camera megapixels

px_height Pixel resolution height

px_width Pixel resolution width

ram RAM in MB

sc_h Screen height in cm

sc_w Screen width in cm

talk_time Maximum talk time in hours

three_g Has 3G or not (1 = Yes, 0 = No)

touch_screen Has touch screen or not (1 = Yes, 0 = No)

wifi Has WiFi or not (1 = Yes, 0 = No)

price_range Target variable (0 = Low, 1 = Medium, 2 = High, 3 = Very High)

The dataset does not contain any missing values, ensuring that no imputation techniques are required.

6. Implementation Approach

6.1 Data Preprocessing

Before training the model, the dataset undergoes preprocessing, including splitting the data into
training, validation, and testing sets. Standardization is applied to numerical features, categorical
values are encoded, and irrelevant features that do not contribute significantly to price classification
are eliminated.

6.2 Model Selection and Training

Several machine learning models are considered for this classification problem, including Logistic
Regression, Support Vector Machine (SVM), Random Forest Classifier, and Gradient Boosting. The
model is trained using the training dataset and tested using the test dataset to evaluate its
performance.

6.3 Model Evaluation

The trained models are assessed using various metrics, including accuracy score to measure prediction
accuracy, precision and recall to evaluate class-wise performance, and confusion matrix to visualize
classification results.

6.4 Optimization and Fine-Tuning

To improve model performance, hyperparameter tuning is performed by adjusting factors such as

6.6 Model Summary and Performance

A multiple linear regression model was trained to predict cellphone prices based on their
specifications.

• Data Preprocessing: The dataset contained cellphone specifications and prices. Key
preprocessing steps included removing the 'Product_id' column, splitting the data into training
and test sets, and scaling the features using StandardScaler.

• Model Training & Evaluation: The multiple linear regression model was trained, and the
model's parameters were obtained by minimizing the mean squared error on the training data.
The performance was evaluated on the test set.

• Performance Metrics:

o R² Score: 0.9387

o Root Mean Squared Error (RMSE): 140.1754

• Visualizations: Various plots were used to visualize the model's performance, including
predicted vs. actual prices and residual analysis.

• This model effectively predicts cellphone prices with high accuracy. However, future
improvements could explore non-linear relationships or additional features to enhance
performance.
6.6 Code

6.7 Snapshots
7. Conclusion

This project successfully demonstrated the application of machine learning in predicting mobile phone
price categories based on technical specifications. The trained models provided reliable classification
results, highlighting key factors influencing phone pricing. By utilizing multiple machine learning
algorithms, we identified the most effective approach for price prediction, achieving high accuracy in
classification. The insights derived from this study can benefit consumers, manufacturers, and retailers
in making informed decisions regarding mobile phone pricing strategies. Future improvements can
include incorporating additional features, leveraging deep learning techniques, and integrating market
trends to further enhance price prediction accuracy.

8. Future Scope

The study of mobile phone price prediction using machine learning has significant potential for future
advancements. Future models can consider the impact of brand reputation on pricing. Expanding the
feature set by incorporating build quality, customer reviews, and software optimizations can improve
prediction accuracy. A real-time price estimation application can be developed, allowing users to input
phone features and receive an estimated price range. Future models can also incorporate market
demand, inflation, and supply chain data for better accuracy. Neural networks and deep learning
techniques can further enhance classification performance, making price predictions more precise.
9. References

1. Bishop, C. M. (2006). Pattern Recognition and Machine Learning. Springer

2. James, G., Witten, D., Hastie, T., & Tibshirani, R. (2013). An Introduction to Statistical Learning.
Springer

3. Hastie, T., Tibshirani, R., & Friedman, J. (2009). The Elements of Statistical Learning: Data
Mining, Inference, and Prediction. Springer

4. Kaggle (2023). Mobile Price Classification Dataset. Retrieved from www.kaggle.com

5. Scikit-learn Documentation (2024). Machine Learning in Python. Retrieved from https://scikit-