Shows and Television

class: center, middle, inverse, title-slide

# Shows and Television
### Zach Mullis, Ben Tibbits
### Last compiled: Apr 27, 2021

---

# Introduction

* With streaming services on the rise its getting more difficult to pick between them and too expensive to pay for all of them

* There are numerous articles discussing which streaming service is best, but we found no studies using statistical methods comparable to ours

* Research Question: What is the best streaming service?
  * Four contenders: Netflix, Hulu, Prime Video, and Disney+
  * Determined by XC scoring system of three criteria:
    1. Amount of content compared to price
    2. Percent of exclusive content
    3. Quality of content

---

# Media1 Data

```
## # A tibble: 4,302 x 5
##    Title                              Year Age    IMDb    RT
##    <chr>                             <dbl> <fct> <dbl> <dbl>
##  1 Inception                          2010 13+     8.8   8.7
##  2 The Matrix                         1999 18+     8.7   8.7
##  3 Avengers: Infinity War             2018 13+     8.5   8.4
##  4 Back to the Future                 1985 7+      8.5   9.6
##  5 The Good, the Bad and the Ugly     1966 18+     8.8   9.7
##  6 Spider-Man: Into the Spider-Verse  2018 7+      8.4   9.7
##  7 The Pianist                        2002 18+     8.5   9.5
##  8 Django Unchained                   2012 18+     8.4   8.7
##  9 Raiders of the Lost Ark            1981 7+      8.4   9.5
## 10 Inglourious Basterds               2009 18+     8.3   8.9
## # … with 4,292 more rows
```

* This is a tibble of our media1 data using the first 5 entries. We aggregated this data using previously developed datasets published on Kaggle.com using a Public Domain or Creative Commons license. After omitting NAs we had a sample size of 4,302. We also averaged ratings between IMDb and Rotten Tomatoes which we called Score.

---

# Media5 Data

```
## # A tibble: 10 x 8
##    Title                              Year Age    IMDb    RT Score Type  Service
##    <chr>                             <dbl> <fct> <dbl> <dbl> <dbl> <fct> <fct>  
##  1 Inception                          2010 13+     8.8   8.7  8.75 Movie N---   
##  2 The Matrix                         1999 18+     8.7   8.7  8.7  Movie N---   
##  3 Avengers: Infinity War             2018 13+     8.5   8.4  8.45 Movie N---   
##  4 Back to the Future                 1985 7+      8.5   9.6  9.05 Movie N---   
##  5 The Good, the Bad and the Ugly     1966 18+     8.8   9.7  9.25 Movie N--P-  
##  6 Spider-Man: Into the Spider-Verse  2018 7+      8.4   9.7  9.05 Movie N---   
##  7 The Pianist                        2002 18+     8.5   9.5  9    Movie N--P-  
##  8 Django Unchained                   2012 18+     8.4   8.7  8.55 Movie N---   
##  9 Raiders of the Lost Ark            1981 7+      8.4   9.5  8.95 Movie N---   
## 10 Inglourious Basterds               2009 18+     8.3   8.9  8.6  Movie N---
```
* We then created `Service` by writing a function creatively named `makeservice` that essentially combined the platform columns from the original data.

---

### Regression Tree

<div class="figure">
<img src="RT.png" alt="Score ~ . - RT - IMDb, media_NT" width="100%" />
<p class="caption">Score ~ . - RT - IMDb, media_NT</p>
</div>

---

# Anova that we can't use

```r
trmod <- aov(Score ~. - RT - IMDb, media_NT)
anova(trmod)
```

```
## Analysis of Variance Table
## 
## Response: Score
##               Df  Sum Sq Mean Sq  F value    Pr(>F)    
## Year           1     0.1    0.10   0.0384  0.844648    
## Age            4   448.4  112.11  42.2577 < 2.2e-16 ***
## Netflix        1    68.3   68.33  25.7559 4.038e-07 ***
## Hulu           1   259.6  259.62  97.8598 < 2.2e-16 ***
## Prime.Video    1    77.5   77.53  29.2227 6.802e-08 ***
## Disney         1    23.0   23.00   8.6684  0.003255 ** 
## Type           1   839.9  839.86 316.5702 < 2.2e-16 ***
## Residuals   4291 11384.0    2.65                       
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
```

---

# QQ-Plot

```r
qqnorm(media1$Score, main = "Score")
qqline(media1$Score)
```

![](Presentation_files/figure-html/unnamed-chunk-4-1.png)

---

# Shapiro Wilks

```r
shapiro.test(media1$Score[1:5000])
```

```
## 
## 	Shapiro-Wilk normality test
## 
## data:  media1$Score[1:5000]
## W = 0.95355, p-value < 2.2e-16
```

* Given that the `$H_0$` for the Shapiro-Wilks test is that the distribution is normal, and our p-value is far below a reasonable `$\alpha$` level, we reject the null. There is evidence that the data is non-normal.

---

# Kruskal-Wallis

```
## 
## 	Kruskal-Wallis rank sum test
## 
## data:  Score by Service
## Kruskal-Wallis chi-squared = 226.3, df = 12, p-value < 2.2e-16
```

`$$H_0: \mu_1 = \mu_2$$`
`$$H_A: \mu_1 \neq \mu_2$$`

* With a p-value far below a reasonable `$\alpha$`, we reject the Null hypothesis. There is evidence that the location parameters ( `$\mu$` ) in `Score` and `Service` have at least one difference.

---

# Amount of content compared to price

# Streaming Data

```
##      Services Price Content       C/P Exclusives       E/C    Score
## 1     Netflix  8.99    5491  610.7898       4936 0.8989255 6.759397
## 2        Hulu 11.99    2657  221.6013       2091 0.7869778 7.211160
## 3 Prime Video  8.99   14498 1612.6808      13647 0.9413022 6.217407
## 4     Disney+  7.99     744   93.1164        688 0.9247312 6.483876
```
* Here we break it down by Platform

---

# Streaming Scatterplot

![](Presentation_files/figure-html/unnamed-chunk-9-1.png)

---

# Percent of exclusive content

![](Presentation_files/figure-html/unnamed-chunk-10-1.png)

---

# Quality of content

![](Presentation_files/figure-html/unnamed-chunk-11-1.png)

---

# Results

```
##      Services C.P E.C Quality Final
## 1     Netflix   2   3       2     7
## 2        Hulu   3   4       1     8
## 3 Prime Video   1   1       4     6
## 4     Disney+   4   2       3     9
```

#### 1st: Prime Video
#### 2nd: Netflix
#### 3rd: Hulu
#### 4th: Disney+

---

# Thanks! Any questions?