As a data science enthusiast, I am constantly seeking out new challenges and opportunities to learn and grow. You can learn about the projects I have completed in the past on this site.

Welcome to my portfolio website!

Patrick Gelsinger

Chief Executive Officer of Intel Corporation

#### **Introduction**

The purpose of this project is to create a model that can predict the chance of a student being admitted to a university based on their application data. This information can be useful for universities in making admission decisions and for students in understanding their chances of being accepted to a particular school.

The data used in this project consists of student admissions data from a university, including the following features:

*   GRE Scores (out of 340)

*   TOEFL Scores (out of 120)

*   University Rating (out of 5)

*   Statement of Purpose and Letter of Recommendation Strength (out of 5)

*   Undergraduate GPA (out of 10)

*   Research Experience (either 0 or 1)

The target variable is the chance of admit, which ranges from 0 to 1.

Kaggle dataset link: <https://www.kaggle.com/datasets/mohansacharya/graduate-admissions>

#### **Data preprocessing**

The data was preprocessed to handle missing values, scale numerical features, and encode categorical features using one-hot encoding.

![](https://preview--smiling-olive-8e2dd.stackbit.dev/\_static/app-assets/images/Screenshot%202022-12-27%20at%2023-48-10%20Google%20Colaboratory-60b3d307.png)![](/images/Screenshot%202022-12-27%20at%2023-48-47%20Google%20Colaboratory-55aab62c.png)

#### **Model training and evaluation**

For this project, linear regression was chosen as the model for prediction. The model was trained using 80% of the data and evaluated on the remaining 20%. The metrics used to evaluate the model were mean absolute error, mean squared error, and R2 score.

The model achieved a mean absolute error of 0.05, a mean squared error of 0.005, and an R2 score of 0.82 on the test set. A scatter plot was also plotted to visualize the model's predictions against the true values.

![](/images/Screenshot%202022-12-29%20at%2001-20-56%20Google%20Colaboratory-913b75d9.png)

![](https://preview--smiling-olive-8e2dd.stackbit.dev/\_static/app-assets/images/Screenshot%202022-12-29%20at%2001-23-08%20Google%20Colaboratory-43899c98.png)

#### **Conclusion**

The linear regression model performed well on the test set with an R2 score of 0.82. Future work could include exploring alternative classification algorithms or fine-tuning the model's hyperparameters to further improve performance. This model can be used to predict the chance of admission to a graduate program based on the input variables as shown below.

![](/images/Screenshot%202023-01-03%20at%2018-33-01%20Google%20Colaboratory-1786824a.png)


Admission Prediction

#### Introduction

Spam emails and SMS messages are a common nuisance for many people, clogging up inboxes and taking up valuable time to sort through. In this project, we aim to build a machine learning model that can accurately classify SMS messages as spam or non-spam (also known as "ham").

To achieve this goal, we will be using the SMS Spam Collection data set, which contains a set of 5,574 SMS messages that have been labeled as either spam or ham. We will use a Naive Bayes classifier to train and test the model on this data. Our evaluation metrics will include accuracy, precision, and recall to understand the model's performance in more detail.

The results of this project have the potential to be applied in a variety of contexts, such as creating spam filters for email or SMS systems, or helping individuals and organizations more efficiently manage their communications.

UCI dataset link : <https://archive.ics.uci.edu/ml/datasets/SMS+Spam+Collection>

#### Data Preprocessing

The SMS Spam Collection data set was loaded into a Pandas DataFrame, and basic statistics and visualizations were used to understand the distribution and characteristics of the data. The messages were then converted into a numerical representation using the CountVectorizer from scikit-learn. To ensure a fair evaluation of the model, the data was split into training and test sets using the train_test_split function from scikit-learn. The training set was used to fit the model, and the test set was used to evaluate the model's performance.

![](/images/Screenshot%202022-12-29%20at%2017-45-34%20Google%20Colaboratory.png)

![](/images/Screenshot%202022-12-29%20at%2017-47-13%20Google%20Colaboratory.png)

![](/images/Screenshot%202022-12-29%20at%2017-48-11%20Google%20Colaboratory.png)

#### Model Training and Evaluation

The Naive Bayes classifier was trained on the training data using the fit method. The classifier was then used to make predictions on the test data, and the performance was evaluated using the classification_report function from scikit-learn.

The evaluation results showed that the model achieved an overall accuracy of 98.4%, with a precision of 99% for detecting spam and a recall of 100% for correctly identifying non-spam messages.

![](/images/Screenshot%202022-12-29%20at%2017-50-55%20Google%20Colaboratory.png)

![](/images/Screenshot%202022-12-29%20at%2017-51-24%20Google%20Colaboratory.png)

#### Conclusion

In this project, we successfully built a machine learning model that can accurately classify SMS messages as spam or non-spam. The model achieved high levels of accuracy, precision, and recall, indicating that it is effective at identifying spam messages.

An example of a spam message is shown below

![](/images/Screenshot%202022-12-29%20at%2017-54-53%20Google%20Colaboratory-dc685232.png)


Welcome to my portfolio website!

Recent Projects

Admission Prediction

Spam Detection

Home

Welcome to my portfolio website!

Recent Projects

Admission Prediction

Spam Detection