Linear Regression. In this section brief description of the linear regression method is presented. Based on this method we will estimate the steel demand in Norway for year 2010. We would like to highlight that Linear Trend Equation and Least Square Method are alternative names for this method. Least Square Method is a powerful technique used to make forecasts when the data represent a linear trend. It determines which line best fits the historical data by minimizing sum of squared deviations around the line. According to Xxxxxxx (1989), the relationship between x and y is given by equation: _
Linear Regression. 3.6.1 Crude Linear Regression Variable Crude Models Beta 95% CI p-value Not Natural Flooring Material (log TTC) = α +β1FLOOR -0.16 -0.60, 0.28 0.4842 Own Animals (log TTC) = α +β1OWNAN 0.19 -0.04. 0.42 0.1046 Storage Container Narrow Mouth (log TTC) = α +β1MOUTH -0.74 -1.37, -0.12 0.0204* Storage Container Covered (log TTC) = α +β1COVERED -0.01 -0.28, 0.25 0.9203 Handwashing Materials Available (log TTC) = α +β1HW_MAT 0.21 -0.08, 0.50 0.1475 Handwashing Frequency (log TTC) = α +β1HW_FRQ 0.07 -0.03, 0.17 0.1913 Improved Toilet Facility (log TTC) = α +β1TOILET_T -0.29 -0.56, -0.01 0.0401* Toilet Facility Shared (log TTC) = α +β1TOILET_S -0.10 -0.38, 0.18 0.4742 Drinking Water Served by Pour (log TTC) = α +β1WTR_SEV -0.10 -0.48, 0.28 0.6193 Improved Water Source (log TTC) = α +β1SAM_SC -1.07 -1.37, -0.78 <.0001* Time Since Collection (log TTC) = α +β1CO_HRS 0.001 -0.002, 0.004 0.5987 Fetching Distance >=500 meters (log TTC) = α +β1FETCH_D -0.26 -0.51, -0.02 0.0312* Water Fetching Time >=30 mins (log TTC) = α +β1FETCH_T -0.34 -0.58, -0.10 0.0053* Water Purchased (log TTC) = α +β1BUY_CO -0.35 -0.63, -0.07 0.0138* Water Treated (log TTC) = α +β1SAM_TR 0.19 -0.35, 0.73 0.4846 SES Quartiles (log TTC) = α +β1SES_Q 0.09 0.02, 0.16 0.0146* Rainy Season (log TTC) = α +β1SEASON 0.61 0.38, 0.84 <.0001* Province (log TTC) = α +β1PROVINCE 0.08 -0.01, 0.17 0.0955 Table 8: Estimates from Crude Linear Regression Models *statistically significant p-value Crude linear regression models were run for each of the predictors as well as for the potential effect modifiers, this information is displayed in Table 8. The variables that had a statistically significant relationship with water quality were storage containers with narrow mouths (p-value=0.0204), improved household toilet facility (p- value=0.0407), improved water source (p-value<0.0001), water fetching distance more or equal to 500 meters (p-value = 0.0312), water fetching time more or equal to 30 minutes(p-value=0.0053), purchasing water (p-value=0.0138), SES quartile status (p- value= 0.0146), and rainy season (p-value <0.0001). The factors that contributed to lower levels of TTC concentration are non-natural flooring materials, narrow mouthed storage containers, covered storage containers, improved toilet facilities, shared toilet facilities, serving water by pouring it, improved water source, fetching distance more or equal to 500 meters, fetching time more or equal to 30 minutes, and purchased water. The factors contr...
