 # Essay On Rainfall Forecasting

Satisfactory Essays
Data Collection and Feature Extraction- The datasets which are required for the rainfall forecasting are downloaded from Indian Meteorological Data (IMD), Mumbai using the Met Data tool. Met data tool provides the dataset for previous 3 years. Met Data tool provides various parameters but only the parameters which affect the rainfall are considered. The parameters like Temperature, Cloud cover, Vapour Pressure, Wet day frequency and rainfall. These parameters are taken as the input parameters. 1. Temperature – The temperature widely affects the precipitation occurring in particular area. The volume of rain that falls into heavy showers depends upon the amount of water vapour. At higher temperature, the atmosphere con1tains more amount…show more content…
Generalized linear models were constructed by John Nelder and Robert Wedderburn. They created the model so that the integration of various statistical models like Linear Regression Model, Poisson model, Logistical model, etc. can be done. The generalized linear model (GLM) is the generalization of the simple linear regression model and the response variable is not normal, it allows other error distribution models. The GLM generalizes the linear regression model by allowing the linear model to be related to the response variable by the link function. In these model, the response variable yi is assumed of the exponential family with mean µi, which is assumed to be some (non-linear) function of xiTβ. Some call these as the non-linear function because mean µi is non-linear with the covariates but Nelder and McCullagh considered them as linear. This is because the covariates affects the distribution of yi only through the linear combination of xiTβ. The Generalized Regression model (GLM) are the broad class of models for like linear regression, logistic regression, ANOVA, log-linear models, etc. There are three components to any GLM- 1. Random component – It refers to conditional distribution of response variable Yi, given the variables- • Conventionally the random variable is the component of the “exponential family” which is the normal (Gaussian),