Crop residues improve soil fertility and health while lessening the soil's vulnerability to erosion in submountainous regions. From 2016 to 2018, a randomized complete block design (RCBD) experiment was carried out. There were 18 treatments made from the residue types of wheat straw (WR) and maize stover (MR), applied at 0, 4, and 6 Mg ha-1 . Wheat and maize crops were planted alternately. The combined data from 2016–17 and 2017–18 demonstrated that crop residue increased wheat grain yield over control by 74% and 83%, respectively, at 4 and 6 Mg ha–1 . Similarly, at both 4 and 6 Mg ha-1 of crop residues, the grain yield of maize increased by 25% between 2016 and 2018. Residue types, MR yielded 12% higher maize grain than WR. The four seasons (2016, 2016–17, 2017 and 2017–18) revealed a higher bulk density (BD), lower total nitrogen (TN), and lower organic matter (OM) without residue incorporation. Higher soil water (SW) contents, higher soil OM, and lower soil bulk density (BD) were observed in the fourth season following the harvest of two wheat and two maize crops. The percent increase in OM was 16, TN 50 and a decrease in BD 4.5 in the fourth season postharvest compared to presowing soil properties. A linear model using multiple linear regression (MLR) and a non-linear model artificial neural networks (ANN) were applied using actual crop yield data, soil properties, and meteorological information. The ANN model slightly outperforms the MLR model in accuracy and explaining yield variation. Thus, the ANN model is more reliable for predicting crop yields based on the data provided. Adding crop residues increases crop yield by improving soil properties.