Dataset for "Predicting the Finished Fabric Width and Areal Density (Grams per Square Meter) of Commercially Produced Plain Single Jersey (100% Cotton) Knitted Fabric using Fuzzy Inference System (FIS)"
Description
This dataset contains the data used in the research titled “Predicting the Finished Fabric Width and Areal Density (Grams per Square Meter) of Commercially Produced Plain Single Jersey (100% Cotton) Knitted Fabric using Fuzzy Inference System (FIS)”. The data was collected from mass production in an industrial environment and consists of key input parameters and corresponding fabric characteristics. The dataset includes the following files: 1) Raw_Data.xlsx – The original dataset containing more than 30,000 observations collected from commercial production of 100% cotton plain single jersey knitted fabric. Each record includes Stitch Length (SL), Yarn Count (YC), Machine Diameter (D), Finished Fabric Width (FW), and Areal Density (GSM). 2) Training_Data.xlsx – A subset of the raw data used for building and training the fuzzy inference system (FIS) model. This dataset includes input parameters (SL, YC, D) and corresponding outputs (FW, GSM). The Subset was created using the "random" library of python. 3) Testing_Data.xlsx – Data used to test and validate the accuracy of the trained fuzzy model. This file contains unseen data points (not included in training) with the same input and output parameters. The raw data was separated into Training Data and Testing Data using python "random" library. 4) 55_Developed_Fabric_Data.xlsx – Real-world experimental observations collected from controlled production runs, specifically designed for model verification and reliability assessment. Purpose of the Dataset The dataset supports research on predicting critical fabric properties (Finished Width and Areal Density) using fuzzy logic-based modeling. It enables further studies, replication of results, and development of predictive tools in the textile manufacturing domain. Key Variables: SL: Stitch Length (for 10Loops in mm) YC: Yarn Fineness (Ne) D: Machine Diameter (inch) FW: Finished Fabric Width (inch) GSM: Areal Density (GSM) Data Format: Excel (.xlsx) Number of Records: Raw Data: ~30,000 observations Training Data: Random subsets of the raw data (~24,000) Testing Data: Random subsets of the raw data (~7,000) Experimental Data: Independent validation dataset (55 Dataset)
Files
Steps to reproduce
To reproduce the results from the study “Predicting the Finished Fabric Width and Areal Density (Grams per Square Meter) of Commercially Produced Plain Single Jersey (100% Cotton) Knitted Fabric using Fuzzy Inference System (FIS)”, follow these steps: 1. Download the Dataset Files Raw_Data.xlsx Training_Data.xlsx Testing_Data.xlsx 55_Developed_Fabric_Data.xlsx 2. Prepare the Environment MATLAB (R2021a or later recommended) with Fuzzy Logic Toolbox. Create Training_Data.xlsx and Testing_Data.xlsx subsets using python. Or use any other method. 3. Load and Inspect Data In MATLAB, import Training_Data.xlsx and Testing_Data.xlsx using readtable or Import Tool. Inputs: SL (Stitch Length for 10 Loops in mm) YC (Yarn Count, Ne) D (Machine Diameter in inches) Outputs: FW (Finished Fabric Width in inches) GSM (Areal Density in GSM) 4. Build the Fuzzy Inference System (FIS) Create a Mamdani-type 1 FIS in MATLAB. Define: 3 input variables (SL, YC, D) with appropriate ranges and membership functions. 2 output variables (FW, GSM) with membership functions. 5. Generate Rules Using Wang & Mendel Algorithm Apply Wang and Mendel’s 5-step rule generation process on the Training_Data.xlsx: Fuzzify each input and output variable into corresponding linguistic terms. For data points, determine the degree of membership for each input/output in its fuzzy set. Create a fuzzy rule from data points using the highest membership values. Assign a degree of certainty (weight) to each rule based on the membership degrees. Eliminate conflicting rules by keeping the one with the highest certainty degree. Finalize the fuzzy rule base. 6. Validate the Model Apply the FIS to Testing_Data.xlsx. Calculate performance metrics: R² (Coefficient of Determination) RMSE (Root Mean Square Error) MAPE (Mean Absolute Percentage Error) 7. Verify with Experimental Data Load experimental_data.xlsx and predict FW and GSM using the trained FIS. Compare predicted vs actual values and compute metrics again.
Institutions
- Bangladesh University of Textiles