Prediction of completion time of an application using decision tree regression model

Abstract

Efficiently estimating the completion time of applications is crucial for project planning, resource allocation, and meeting stakeholders' expectations. In this study, we propose a predictive model using the Decision Tree Regression (DTR) algorithm to accurately forecast the completion time of software applications during the development lifecycle. The dataset used for training and testing the model comprises historical records of completed projects, containing various application characteristics and their corresponding actual completion times. First, we preprocess the dataset to handle missing values, normalize numerical features, and encode categorical variables appropriately. We then employ the Decision Tree Regression algorithm to build a predictive model by recursively partitioning the data based on feature thresholds, resulting in a tree-like structure. The DTR model offers interpretability, which allows us to gain insights into the important factors influencing the completion time. To evaluate the model's performance, we adopt various metrics, including Mean Absolute Error (MAE), Mean Squared Error (MSE), and R-squared (R2) score. Additionally, we compare the DTR model's performance with other regression models, such as Linear Regression and Random Forest Regression, to showcase its effectiveness in handling application completion time prediction. The experimental results demonstrate that the proposed Decision Tree Regression model outperforms other traditional regression methods in accurately forecasting the completion time of applications. Moreover, the model's interpretability aids project managers in understanding the underlying relationships between application features and completion time, facilitating better decision-making and resource management. In conclusion, our study presents a robust approach to predict the completion time of software applications using the Decision Tree Regression model. The proposed methodology can significantly benefit software development projects, providing stakeholders with reliable estimates for better planning and execution, ultimately leading to improved project success rates.