Design and Analysis of Urban Land Lease Price Predicting Model Using Batch Gradient Descent Algorithm

Standard and econometric models are appropriate for causal relationships interpretations among facets of the economy. But with prediction, they tend to over-fit samples simplify poorly new, undetected data. This paper presents a batch gradient algorithm predicting rice land large datasets. uses descent minimize cost function, iteratively possible combinations number iterations i=1500 learning rates, 0.01, 0.02, 0.03 linear regression case i = 100, 0.3, 0.2, 0.1 multiple case. The Octave-4.0.3(GUI) implementing 129 lease bid price Mekelle City as training sets feature inputs two three regressions. Using best fitting parameters found by dataset J=67.82. model predicts an accuracy 92.6% using LR 91.15% MLR 315 m2 size. As rate increases, increase decrease respectively equal but model’s prediction error increments slowly. With regression, lowers, under fits drastically (with 60%) 91.5% ordinary equations. So, equations provides fit