Data Visualization

Row

Barplot (Global)

Barplot (US Market)

Row

Publisher Summary

Publisher Name Total_Clicks Avg_Cost_Click Total_Revenue Cost Net_Revenue ROA Net_Rev_Click
KAYAK 2839 1.2564741 233694.0 3567.13 230126.9 64.513172 81.059130
Yahoo - US 45598 1.9988757 882288.9 46197.82 836091.1 18.098063 18.336136
MSN - Global 11217 2.1529982 145524.2 12160.36 133363.9 10.967098 11.889444
MSN - US 10808 2.8674701 181549.8 16098.49 165451.3 10.277445 15.308227
Google - Global 72895 2.2249594 929549.8 120946.71 808603.1 6.685614 11.092710
Overture - Global 60899 0.8047588 430084.7 64295.86 365788.8 5.689151 6.006483
Google - US 192109 2.3839418 1745064.4 353640.60 1391423.9 3.934570 7.242887
Overture - US 119323 0.7639206 347433.2 141976.07 205457.2 1.447125 1.721857

Keyword Summary

Keyword Total_Clicks Avg_Cost_Click Total_Revenue Cost Net_Revenue ROA Net_Rev_Click
low international airfare 2 0.66 1574.20 1.31 1572.89 1198.39 786.44
[airfrance] 59 0.04 1665.15 2.31 1662.84 719.06 28.18
air france online booking 8 0.28 935.00 2.20 932.80 424.00 116.60
air travel to paris 13 0.40 1513.00 5.21 1507.79 289.26 115.98
fly to florence 29 1.57 8777.95 34.71 8743.24 251.88 301.49
air discount france ticket 5 0.36 390.15 1.69 388.46 230.20 77.69
cheap fare to paris 41 0.86 7418.80 36.32 7382.48 203.23 180.06
airfrance,com 90 0.14 2535.55 12.56 2522.99 200.83 28.03
airfrance website 10 0.99 1815.60 9.94 1805.66 181.70 180.57
air france us 328 0.40 24279.40 141.59 24137.81 170.48 73.59

Publisher Summary (US Market)

Publisher Name Total_Clicks Avg_Cost_Click Total_Revenue Cost Net_Revenue ROA Net_Rev_Click
Google - US 6645 2.583143 60149.4 16774.212 43375.188 2.5858256 6.5274925
Yahoo - US 2475 2.185422 8115.8 6470.962 1644.838 0.2541875 0.6645808
MSN - US 1 2.675000 0.0 2.675 -2.675 -1.0000000 -2.6750000

Keyword Summary (US Market)

Keyword Total_Clicks Avg_Cost_Click Total_Revenue Cost Net_Revenue ROA Net_Rev_Click
low international airfare 2 0.66 1574.20 1.31 1572.89 1198.39 786.44
france air ticket 16 0.93 2706.40 22.45 2683.95 119.55 167.75
france airline ticket 6 1.13 494.70 6.68 488.02 73.11 81.34
florence cheap tickets 5 1.91 490.45 9.70 480.75 49.56 96.15
international airfares 6 3.48 923.95 20.40 903.55 44.29 150.59
france airfare sale 5 2.78 470.05 13.94 456.11 32.73 91.22
paris cheap airline 24 1.96 1685.55 51.19 1634.36 31.93 68.10
france airlines 152 2.43 11301.60 343.35 10958.25 31.92 72.09
france air flight 80 0.77 2100.35 67.41 2032.94 30.16 25.41
paris cheap air tickets 7 2.58 494.70 17.42 477.28 27.39 68.18

Logistic Model

Column

Model Summary (Without Rescaling)


Call:
glm(formula = ROA_Binary ~ `Search Engine Bid` + Clicks + Impressions + 
    `Engine Click Thru %` + `Avg. Pos.` + `Total Cost/ Trans.` + 
    `Total Cost`, family = "binomial", data = training_data)

Deviance Residuals: 
    Min       1Q   Median       3Q      Max  
-5.9252  -0.3612  -0.3295  -0.2854   2.8769  

Coefficients:
                        Estimate Std. Error z value Pr(>|z|)    
(Intercept)           -1.871e+00  2.447e-01  -7.646 2.07e-14 ***
`Search Engine Bid`   -6.175e-02  2.290e-02  -2.697 0.007006 ** 
Clicks                 3.427e-03  4.842e-04   7.077 1.47e-12 ***
Impressions           -2.386e-06  6.330e-07  -3.770 0.000163 ***
`Engine Click Thru %` -7.832e-03  4.601e-03  -1.702 0.088736 .  
`Avg. Pos.`           -3.396e-01  9.404e-02  -3.611 0.000305 ***
`Total Cost/ Trans.`   2.023e-03  4.271e-04   4.735 2.19e-06 ***
`Total Cost`          -9.400e-04  2.142e-04  -4.389 1.14e-05 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 1858.1  on 3606  degrees of freedom
Residual deviance: 1568.8  on 3599  degrees of freedom
  (1 observation deleted due to missingness)
AIC: 1584.8

Number of Fisher Scoring iterations: 7

Model Summary (Normalized)


Call:
glm(formula = ROA_Binary ~ search_engine_bid_norm + clicks_norm + 
    impressions_norm + engine_click_thru_norm + avg_pos_norm + 
    total_cost_by_trans_norm + total_cost_norm, family = "binomial", 
    data = training_data)

Deviance Residuals: 
    Min       1Q   Median       3Q      Max  
-5.9252  -0.3612  -0.3295  -0.2854   2.8769  

Coefficients:
                         Estimate Std. Error z value Pr(>|z|)    
(Intercept)               -1.8712     0.2447  -7.646 2.07e-14 ***
search_engine_bid_norm    -1.6981     0.6297  -2.697 0.007006 ** 
clicks_norm              116.5625    16.4695   7.077 1.47e-12 ***
impressions_norm         -19.9094     5.2805  -3.770 0.000163 ***
engine_click_thru_norm    -1.5664     0.9203  -1.702 0.088736 .  
avg_pos_norm              -5.0933     1.4106  -3.611 0.000305 ***
total_cost_by_trans_norm  19.4104     4.0990   4.735 2.19e-06 ***
total_cost_norm          -43.4189     9.8921  -4.389 1.14e-05 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 1858.1  on 3606  degrees of freedom
Residual deviance: 1568.8  on 3599  degrees of freedom
  (1 observation deleted due to missingness)
AIC: 1584.8

Number of Fisher Scoring iterations: 7

Column

Confusion Matrix

Confusion Matrix and Statistics

          Reference
Prediction   0   1
         0 832  54
         1   5  11
                                          
               Accuracy : 0.9346          
                 95% CI : (0.9164, 0.9498)
    No Information Rate : 0.9279          
    P-Value [Acc > NIR] : 0.2423          
                                          
                  Kappa : 0.2503          
                                          
 Mcnemar's Test P-Value : 4.129e-10       
                                          
            Sensitivity : 0.9940          
            Specificity : 0.1692          
         Pos Pred Value : 0.9391          
         Neg Pred Value : 0.6875          
             Prevalence : 0.9279          
         Detection Rate : 0.9224          
   Detection Prevalence : 0.9823          
      Balanced Accuracy : 0.5816          
                                          
       'Positive' Class : 0

ROC and AUC Graph

GINI Tree Model

Column

GINI Tree Figure

Column

Plot CP Graph

ROC and AUC Graph