Principles of Digital Analytics

Principles of Digital Analytics

  • Analytics is used in the extraction of useful information from data, where the main aim is in supporting decision making, whether the information is considered to be strategic or operational in nature. Descriptive analytics describes what happened in the past. Descriptive analytics are meaningful since they allow people to learn from past occurrences and understand how they might impact future outcomes. In general, descriptive analytics are pre-canned and operational reports, MIS and dashboards and among others. On the other hand, predictive analytics is used in determining what could happen into the future. Often they require larger data set tools and expertise. For instance, the channels that are likely to perform better in the next season based on the data about the performances of the past seasons.
  1. (a)

Given that GNP growth is exponential in nature and further assume that unemployment rate and inflation can linearly predict the GNP growth.

The model will then be constructed in the following way:

GNP growth = α + β1 Unemployment Rate + β2 Inflation + μ

 

Model Summary
Model R R Square Adjusted R Square Std. Error of the Estimate
1 .938a .881 .642 .0004279
a. Predictors: (Constant), Inflation, Unemployment_Rate

The value .881 indicates that 88 percent of the variance in GNP growth can be predicted from unemployment rate and inflation. However, this general measure of the strength of the association between GNP growth on the one hand, and inflation and unemployment rate on the other does not reflect the extent to which any of the two (inflation and unemployment rate) is associated with GNP growth.

 

ANOVAa
Model Sum of Squares df Mean Square F Sig.
1 Regression .000 2 .000 3.689 .345b
Residual .000 1 .000    
Total .000 3      
a. Dependent Variable: GNP_GROWTH
b. Predictors: (Constant), Inflation, Unemployment_Rate

The F-value refers to the mean square regression (0.0) divided by the mean square residuals and which leads to the outcome of F = 3.689. the p-value linked with this F value is large (.345). the F-value and F-sig are used in answering question regarding the probability of both inflation and unemployment rate predicting GNP growth. The Sig. is compared is often compared to the alpha level (.05), and since it is larger, the null hypothesis is not accepted and the conclusion is that the inflation and unemployment do not show statistically significant relationship with GNP growth. It is important to note that this is the overall significance test that assessed whether the two (inflation and unemployment) when used together can reliably predict GNP growth, and it does not address the ability of either inflation or unemployment rate predicting GNP growth.  The ability of either unemployment rate or inflation in predicting GNP growth is addressed here in below where each of the two variables are listed.

 

Coefficientsa
Model Unstandardized Coefficients Standardized Coefficients t Sig.
B Std. Error Beta
1 (Constant) -.002 .007   -.335 .794
Unemployment_Rate .001 .001 .406 1.035 .489
Inflation .001 .001 .676 1.726 .334
a. Dependent Variable: GNP_GROWTH

 

 

Coefficientsa
Model Unstandardized Coefficients Standardized Coefficients t Sig. Correlations Collinearity Statistics
B Std. Error Beta Zero-order Partial Part Tolerance VIF
1 (Constant) -.002 .007   -.335 .794          
Unemployment_Rate .001 .001 .406 1.035 .489 .725 .719 .358 .777 1.287
Inflation .001 .001 .676 1.726 .334 .868 .865 .596 .777 1.287
a. Dependent Variable: GNP_GROWTH

The p-value of the coefficient of unemployment rate (p = .489) and coefficient of inflation (p= .334) are all greater than the alpha (α = .05), and which means that the coefficient of the two independent variables are not statistically significant, and which means that individually, inflation and unemployment rate do not predict the GNP growth.

 

 

 

(b)

Collinearity Diagnosticsa
Model Dimension Eigenvalue Condition Index Variance Proportions
(Constant) Unemployment_Rate Inflation
1 1 2.993 1.000 .00 .00 .00
2 .006 22.186 .04 .01 .87
3 .000 84.104 .96 .99 .13
a. Dependent Variable: GNP_GROWTH

Multicollinearity can be examined through the variance proportions columns above or simply my making a conclusion on the largeness of the standard error values. The standard errors between the independent variables are small in the coefficients table, and which communicates minimal collinearity among the variables.

Autocorrelation often deals with time, but in the context of regression, if there is some natural ordering of the data set, it is multicollinearity, it is the rows instead of columns (Zubair and Adenomon 105). Technically the thing here involves the residuals of the regression. This means that they show some sort of pattern across index or time, it affects the ability of the regression to predict well. It also affects parameter estimates and standard errors/ Heteroscedasticity is closely related to autocorrelation. Basically, the residuals of the regression are independently varying (hetero in this case means separate.

Residuals Statisticsa
  Minimum Maximum Mean Std. Deviation N
Predicted Value .009189 .010661 .009842 .0006711 4
Residual -.0003531 .0002146 .0000000 .0002471 4
Std. Predicted Value -.973 1.221 .000 1.000 4
Std. Residual -.825 .501 .000 .577 4
a. Dependent Variable: GNP_GROWTH

(c)

 

Model Summaryb
Model R R Square Adjusted R Square Std. Error of the Estimate Durbin-Watson
1 .790a .625 -.126 .33379 1.558
a. Predictors: (Constant), GNP_GROWTH, Inflation
b. Dependent Variable: Unemployment_Rate

 

The value .625 indicates that 63.5percent of the variance in unemployment rate be predicted from GNP growth and inflation.  The Durbin-Watson test stat of 1.558 measures autocorrelation and since it is less than 2, then it means that there is no autocorrelation in the sample selected.

 

Coefficientsa
Model Unstandardized Coefficients Standardized Coefficients t Sig. Correlations Collinearity Statistics
B Std. Error Beta Zero-order Partial Part Tolerance VIF
1 (Constant) 4.936 2.972   1.661 .345          
Inflation -.511 .992 -.634 -.515 .697 .472 -.458 -.315 .247 4.045
GNP_GROWTH 560.948 541.995 1.275 1.035 .489 .725 .719 .634 .247 4.045
a. Dependent Variable: Unemployement_Rate

 

SUMMARY OUTPUT            
             
Regression Statistics          
Multiple R 0.481081          
R Square 0.231439          
Adjusted R Square 0.154582          
Standard Error 13.03936          
Observations 12          
             
ANOVA            
  df SS MS F Significance F  
Regression 1 512 512 3.011322 0.113339  
Residual 10 1700.25 170.025      
Total 11 2212.25        
             
  Coefficients Standard Error t Stat P-value Lower 95% Upper 95%
Intercept -5.25 27.91561 -0.18807 0.854586 -67.4499 56.94985
Height in feet 8 4.610111 1.735316 0.113339 -2.27197 18.27197

 

Shelf position with a p-value of 0.11 against the α =.05, the null hypothesis that shelf position predicts Oreos sales is not accepted because the coefficient of Shelf position is not statistically significant.

  1. A discrete choice model describes, explains, and predicts choices that exist between at least two discrete alternatives. The choices can be compared with the standard consumption models where the quality of every good consumed is embedded with the assumption of a continuous variable. The ability to make predictive choices between at least two alternatives will always be hard because there is always an opportunity cost(Haghani, Bliemer and Hensher 3). Logistic regression is just a special case of discrete choice modeling. Discrete choice theory simply refers to modeling outcomes that are discrete. Logistic regression simply models a binary outcome. Other models used to look at discrete choice modeling probably include probit regression, multinomial regression, ordered logit, ordered probit, modeling ratings, item response theory models, and probably any machine learning classifier technique.

Businesses revolve around customers because customers generate revenue for a company. The more satisfied the customers are, the more prosperous a business becomes (Niknamian 133). Understanding customer value impacts the sustainability and viability of a firm. For this problem, calculating customer value gives a better understanding of how the relationship will be throughout its involvement with the company, hence being one of the most important and critical challenges a company needs to analyze.

Well, customer value is the estimate of the financial worth a business acquires from its entire relationship with a customer. In short, the sum of returns the company gets over from a customer. Using this metric, the company will know how well are the customer’s resonating with them. The company can easily differentiate the customers having low customer value, mid customer value and high customer value, modify and target their marketing strategies and resources to the right group of people generating more returns. Additionally, the business can keep on calculating its customer value. The increase in the customer value points means that the business is making a better impression on its current customers.

Annual retention for a cell phone subscriber = 70%

Customer generates =$300 annual profit

Annual discount rate = 8%

Customer value =

=

= $552.63

 

Works Cited

Haghani, Milad, Michiel, C, J Bliemer and David, A Hensher. “The landscape of econometric discrete choice modelling research.” Journal of Choice Modelling 40 (2021): 100303.

Niknamian, Sorush. “The Use of Customer value changing trends in business analysis.” Quantitative Economics and Management Studies 1.2 (2020): 130-138.

Zubair, Mohammed, Anono and Monday, Osagie Adenomon. “COMPARISON OF ESTIMATORS EFFICIENCY FOR LINEAR REGRESSIONS WITH JOINT PRESENCE OF AUTOCORRELATION AND MULTICOLLINEARITY.” Science World Journal 16.2 (2021): 103-109.

Related Posts:

 





Is this your assignment or some part of it?

We can do it for you! Click to Order!



Order Now


Translate »

You cannot copy content of this page