Project 1 Report : Logistic Regression
نویسنده
چکیده
In this project, we study learning the Logistic Regression model by gradient ascent and stochastic gradient ascent. Regularization is used to avoid overfitting. Some practical tricks to improve learning are also explored, such as batch-based gradient ascent, data normalization, grid searching, early stopping, and model averaging. We observe the factors that affect the result, and determine these parameters. Finally, we test the algorithm on Gender Recognition [DCT] dataset.. We use LBFGS method on the same dataset as a comparison. Our model trained by stochastic gradient ascent achieves around 92.89% accuracy.
منابع مشابه
Transaction Anomaly Detection CMPS 242 - Project Report
This project uses a series of learning approaches for a transaction anomaly detection problem. The input data is a 19-feature dataset of transaction records with high binary class skewness (1% vs 99%). We apply Logistic Regression, AdaBoost and Boosting Trees and we use Lift and prediction accuracy metrics to evaluate our results. All approaches were based on the material covered in the course ...
متن کاملPredicting Probability of Loan Default Stanford University , CS 229 Project report
Stanford University, CS229 Project report Jitendra Nath Pandey, Maheshwaran Srinivasan 12/15/2011 Abstract: Extending credit to individuals is necessary for markets and societies to function smoothly. Estimating the probability that an individual would default on his/her loan, is useful for banks to decide whether to sanction a loan to the individual and is also useful for borrowers to make bet...
متن کاملUsing logistic regression to predict developer responses to Coverity Scan bug reports
This report presents the state of my ongoing work to create statistical models that can be used to make predictions about how developers will respond to bug reports issued by the Coverity tool in the Open Source Scan project. I present models that can predict the following probabilities for a given report based on properties of the report itself and especially on the development history of the ...
متن کاملAssignment 2 Report
Our project builds a binary classifier that predicts the existence of a connection between any pair of nodes in a facebook ego-net graph. We investigate the most representative features to use and compare the performance of Logistic Regression (LR), Support Vector Machine (SVM), and Random Forest (RF). SVM performs well on the specific task. LR performs decently and will potentially play a larg...
متن کاملRank Aggregation in Scientific Publication Databases Based on Logistic Regression
The goal of the d-Rank project is to study rank aggregation in scientific publication databases. In our work we focus in particular on document ranking in the domain of particle physics and we work with the collection of CERN publications called the CERN Document Server. In this report we present the main advances achieved within the second phase of the project. The most important achievements ...
متن کامل