Knitting is a vital sector of the fabric manufacturing industry. Concurrently, machine learning is emerging as a highly regarded technique for predicting patterns and classifying various parameters derived from datasets. This study aims to establish a comprehensive database of knitted fabrics that encompasses a variety of parameters related to yarn types, machinery, and fabric characteristics. The raw data was collected from a knitting factory, after which the dataset was processed with various pre-processing techniques using domain knowledge and the Python programming. These techniques included data cleaning, normalization, and feature engineering, all of which were crucial in ensuring the quality and usability of the dataset. Drawing on expertise in knitting science, several new parameters were formulated, and specific complex parameters were subsequently deconstructed into two or three distinct components. The finalized dataset has 12569 rows and 38 columns. This article also discusses potential applications of the dataset, such as identifying a polynomial relationship between grams per square meter (GSM) and yarn count for single jersey fabrics, having an R² score of 0.77. Furthermore, a quadratic relationship between the tightness factor and stitch length was observed, with an R² score of 0.78. Among various machine learning models to predict GSM, Random Forest and XGBoost consistently outperformed across all metrics (R² score, Mean Absolute Error, and Mean Square Error).