In [11]:

    

library(dplyr)


    

In [12]:

    

quality_prediction_and_page_views <- read.table("../results/sql_queries/entity_views_and_aggregated_revisions/entity_views_and_aggregated_revisions_and_quality_scoring_prediction_converted_20121101.tsv", header=FALSE, sep="\t")


    

In [13]:

    

colnames(quality_prediction_and_page_views) <- c('entity_id','number_of_revisions', 'page_views', 'prediction', 'ordinal_score')


    

In [14]:

    

head(quality_prediction_and_page_views)


    

    






entity_id
number_of_revisions
page_views
prediction
ordinal_score

	
Q1      
1118    
 5925113
D       
2       
	
Q100    
 491    
17385277
D       
2       
	
Q1000   
 596    
 4903348
D       
2       
	
Q1001   
 973    
16341176
E       
1       
	
Q1002   
 143    
  160651
E       
1       
	
Q1003   
 128    
  820789
E       
1       

In [15]:

    

correlation_output <- cor(quality_prediction_and_page_views$page_views,quality_prediction_and_page_views$ordinal_score, method="spearman")


    

    





0.40610963522853

In [29]:

    

correlation_output


    

    





0.40610963522853

In [16]:

    

quality_prediction_and_page_views_model <- lm(quality_prediction_and_page_views$page_views ~ quality_prediction_and_page_views$ordinal_score)


    

In [17]:

    

summary(quality_prediction_and_page_views_model)


    

    






Call:
lm(formula = quality_prediction_and_page_views$page_views ~ quality_prediction_and_page_views$ordinal_score)

Residuals:
       Min         1Q     Median         3Q        Max 
 -24419800   -3697762   -3591400   -2879327 5643588634 

Coefficients:
                                                 Estimate Std. Error t value
(Intercept)                                     -17022962    5500204  -3.095
quality_prediction_and_page_views$ordinal_score  20721525    4129651   5.018
                                                Pr(>|t|)    
(Intercept)                                      0.00198 ** 
quality_prediction_and_page_views$ordinal_score 5.43e-07 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 119700000 on 4390 degrees of freedom
Multiple R-squared:  0.005703,	Adjusted R-squared:  0.005476 
F-statistic: 25.18 on 1 and 4390 DF,  p-value: 5.435e-07

In [18]:

    

class_a_quality_prediction_and_page_views <- filter(quality_prediction_and_page_views, prediction=="A")


    

In [19]:

    

sorted_ascend_class_a_quality_prediction_and_page_views <- dplyr::arrange(class_a_quality_prediction_and_page_views, page_views)


    

In [20]:

    

head(sorted_ascend_class_a_quality_prediction_and_page_views, n=10)


    

    






entity_id
number_of_revisions
page_views
prediction
ordinal_score

In [ ]:

    




    

In [21]:

    

class_e_quality_prediction_and_page_views <- filter(quality_prediction_and_page_views, prediction=="E")


    

In [22]:

    

sorted_desc_class_e_quality_prediction_and_page_views <- dplyr::arrange(class_e_quality_prediction_and_page_views, desc(page_views))


    

In [23]:

    

head(sorted_desc_class_e_quality_prediction_and_page_views, n=10)


    

    






entity_id
number_of_revisions
page_views
prediction
ordinal_score

	
Q866      
 621      
2079749157
E         
1         
	
Q918      
 438      
2063217449
E         
1         
	
Q1860     
1050      
1419981073
E         
1         
	
Q328      
 643      
 378666601
E         
1         
	
Q2736     
 702      
 270665514
E         
1         
	
Q515      
 617      
 233190433
E         
1         
	
Q2184     
 315      
 226793302
E         
1         
	
Q432      
 562      
  67870906
E         
1         
	
Q2807     
 881      
  64656434
E         
1         
	
Q1065     
 882      
  62315605
E         
1         

In [24]:

    

nrow(class_e_quality_prediction_and_page_views)


    

    





3259

In [25]:

    

nrow(filter(quality_prediction_and_page_views, prediction=="D"))


    

    





1133

In [26]:

    

nrow(filter(quality_prediction_and_page_views, prediction=="C"))


    

    





0

In [27]:

    

nrow(filter(quality_prediction_and_page_views, prediction=="B"))


    

    





0

In [28]:

    

nrow(filter(quality_prediction_and_page_views, prediction=="A"))


    

    





0

In [ ]: