WITHDRAWN: Prediction of fruit quality parameters using peel color in Citrus Reticulata L. fruit by multiple linear regression and artificial neural network approach

Abstract Kinnow mandarin fruit color changes from deep green to completely orange during the period of peak maturity, which indicates if the fruit should be harvested. On the basis of the peel color, fifteen fruits were being selected for harvest, and their colors were measured with the chromameter which measures color in CIE-Lab color space. Green to Orange color fruits were divided into five regions (Green, Green-Yellow, Yellow, Yellow-Orange, Orange). The fruit samples in triplicate were used for all the laboratory analysis. The effects of color parameters i.e. L-value, a-value and b-value in different color regions on chemical parameters i.e. total soluble solids (TSS), titratable acidity (TA), fruit juice content (Juice %), maturity index (MI), ascorbic acid content (AA) and pH value of the fruit were studied using descriptive and correlation analysis. The model approaches i.e. multiple linear regression (MLR) and artificial neural network (multi-layer perceptron) (MLP-ANN) were implemented on the fruit sample data to predict the important fruit quality parameters which were prominent in determining the fruit color. The predicted data was plotted against the actual data using boxplot and scatter plot for both the models, the linearly best fitted and statistical significance was determined using R 2 , RMSE. Green and Green-Yellow colored region fruits have high in titratable acidity, ascorbic acid content and lower soluble sugars, maturity index and pH value. There was a steeper transition visible in TSS, juice content, maturity index, AA and pH value for the color shift from Yellow region to Yellow-Orange region. The better and optimal predictions were made with MLP-ANN for total soluble solids, titratable acidity and fruit juice content. The ANN model could be used in future for prediction of maturity indices of kinnow fruit in different maturity regions as per the consumer and market preferences.