Apple is one of the important horticultural crops of Himachal Pradesh. The present study was conducted to see the apple production trends in Himachal Pradesh. The apple production data for the period 1973-74 to 2017-18 was used to fit various linear and non-linear models to study the apple production trends. The area under apple increased with compound growth rate 2.89% and production increased with compound growth rate 3.67%. The parameters of the various linear and non-linear models were estimated for apple production..Theil?s inequality and Chow test indicated that it was not appropriate to predict the area under apple crop in the state however the quadratic model fitted well among the three linear models to the production data of apple with highest adjusted R2 (0.526) and lowest RMSE (134.216) values. Among the various non-linear models the Rational model was the best fitted model based on various goodness of fit criteria viz., MSE (17105.973), RMSE (130.79), MAE (96.308), and AIC(446.623) values. The assumptions of independence and normality of error terms were examined by the ?Run-test? and Shapiro-Wilk?s test respectively. Durbin Watson test was used to examine autocorrelation among residuals for the various fitted models. From the present analysis of data it was observed that all the models followed the assumptions of linear and non-linear models. On comparing different statistics of analysis of both linear and non-linear models, the non-linear Rational model performed better for describing apple production in Himachal Pradesh. Keywords: Adjusted R2 , MSE, MAE, AIC, Durbin Watson test, Non-linear models.

  Crop yield models are abstract presentation of the interaction of crop with its environment and can range from simple correlation of yield with a finite number of variables to the complex statistical models with predictive end. Autoregressive Integrated Moving Average (ARIMA) models have been fitted for the yield data of sugarcane crop in Karnal, Ambala and Kurukshetra districts of Haryana. The crop yield data of the past four/five decades have been used for the model building and the forecast values are obtained for the years 2016-17 to 2020-21. After experimenting with different lags of the moving average and autoregressive processes; ARIMA (0,1,1) for Karnal, Ambala and Kurukshetra districts have been fitted for sugarcane yield forecasting. The overall results indicates that the percent relative deviations of the forecast yield(s) from the real-time yield(s) are within acceptable limits and favors the use of ARIMA models to get short-term forecast estimates. Keywords: Autocorrelation, Partial autocorrelation, Differencing, Stationarity and invertibility.

  The interplay between computer science and agriculture has led to the collection of huge amounts of information in agricultural datasets. The process of turning low level data into high level knowledge is popularly known as data mining. The field of agriculture has many applications and one important application is in terms of deriving useful patterns like characteristics of disease and varieties. Understanding the distinctions between numerous contrasting groups is a crucial issue in data analysis in order to discover new patterns. These contrasting groups can represent various item classes, such as disease or varieties for different crop groups. Contrast sets are the combinations of attributes and their values that differ meaningfully in their distribution across groups. STUCCO algorithm is a search method for mining contrast sets leading to pattern discovery. The algorithm?s applicability for pattern detection has been demonstrated using agricultural datasets in this paper. Approach resulted in significant pattern discovery for description of soybean disease and IRIS varieties characteristics. Keywords: Contrast set, Pattern discovery, Statistical significance, STUCCO algorithm.

  Balanced treatment incomplete block (BTIB) designs are quite popular for comparing test versus a single control treatment. In this article, we extend the class of BTIB designs by introducing nearly BTIB designs. Nearly BTIB designs can act as a useful alternative to BTIB designs when the latter is not available for a given parametric combination. An algorithm is proposed to construct nearly BTIB designs and a list of such designs is also provided in a practically useful parametric range. Keywords: Nearly balanced treatment incomplete designs, BTIB, Test, Control.

  Additive main effects and multiplicative interaction (AMMI) models have been used to analyze genotype-by-environment interactions (GEI). Applicability of AMMI model depends on the assumption of normally distributed error with a constant variance. However, in case of count data, the appropriateness of AMMI model may be inappropriate as it does not conform to these statistical assumptions. It can be handled by introducing multiplicative terms for interaction in wider class of modeling, Generalized Linear Models known as Generalized AMMI (GAMMI) model. In this paper, GAMMI model with Poisson distribution and log link function has been used to analyze GEI data for number of cobs in maize during Kharif season for 32 genotypes grown over 13 locations across India. Here, Generalized AMMI model replaces the classical AMMI to model GEI in case of count data. For genotype by environment (G?E) interaction in cobs count data, GAMMI model has been applied and compared with the usual AMMI model for original as well as squared root transformed data. The performance of the GAMMI model was compared with the classical AMMI model and the AMMI model applied to transformed (square-root) data. The criterion used for comparison was explained variability in terms of number of axes which were determined by Gollob test. On comparison of results it was found that maximum variation was explained by the first three axes in case of GAMMI model i.e, 63.13 percent. Keywords: GEI, AMMI, Number of cobs, GAMMI, Log link, Poisson distribution.