# Research

### Content Outline

## CEM Research for Data Analysts and Data Scientists

For the curious CEM data analyst or data scientist seeking to mine deeper insights from customer data, our **Research** module offers a swiss army knife of tools.

Every statistical analysis starts with a good clean data set. While it may sound easy, in reality organizing the data takes several days to ensure it is correctly sampled, clean, and valid for analysis. CloudCherry handles all of this automatically, letting CX data analysts work on the implementation of their statistical model instead of the plumbing. They can consume this data in **R** or other tools to build, test and validate further models.

### Research Tools

- Correlation
- Linear Regression
- Logistic Regression
- Significance T-test
- Path Analysis
- Decision Trees⁺
- Random Forest⁺
- K-Means Clustering⁺
- Structural Equation Modeling
- Confirmatory Factor Analysis⁺
- Topic Modeling⁺

⁺ _{through the CloudCherry R SDK}

### Ready integration with R

Our ready integration with **R** opens a range of possibilities for discovering CX insights as well as full flexibility for creating custom reports tailored to branding, layouts, visualization and export preferences.

Reports created using R or R Studio can be automatically updated by synchronization with CloudCherry over secure APIs.

## Use CX Research Tools

### Correlation

To run the Pearson Correlation, select the variables that you want to test the correlations between and click SUBMIT.

### Linear Regression

To run the Linear Regression, first select the independent variables that you want to analyze. Next, choose from the available dependent variables and click SUBMIT. The list of dependent variables is automatically filtered to present a list that are suitable for running a linear regression.

### Logistic Regression

To run the Logistic Regression, first select the independent variables that you want to analyze. Next, choose from the available dependent variables and click SUBMIT. The list of dependent variables is automatically filtered to present a list that are suitable for running a logistic regression.

### Significance T-test

### Path Analysis

### Decision Trees⁺

### Random Forest⁺

### Betasq⁺

### Structural Equation Modeling

### K-Means Clustering⁺

### Factor Analysis

### Topic Modeling

## Export to R

The CloudCherry R SDK lets you import CloudCherry data into R for the purpose of more extensive manipulation using the full range of libraries available in R.

### Installing CloudCherry for R

The package is not yet on CRAN. To install the latest development version you can install from the CloudCherry Github repository. Just follow these steps in the R Console.

```
# latest version from Github
install.packages('devtools')
library(devtools)
install_github('cloudcherry-r-sdk','getcloudcherry')
```

### Example - Themes with Sentiment

Once imported, you can use the full range of R’s toolchain for further analysis. The sample below shows a simple example of manipulating sentiment scores and themes from CloudCherry’s text analytics to look at aggregates of sentiment across different themes.

```
```
Themes = as.vector(df$Theme)
sentiment = as.vector(df$Sentiment)
top_themes = sort(table(df$Theme), decreasing = TRUE)
top_themes = names(top_themes[1:5])
theme_sentiment_split = c()
for (i in top_themes){
df_subset = df[df$Theme == i,]
sentiment_counts = table(df_subset$Sentiment)
str_neg_percent = (as.vector(sentiment_counts["Strong Negative"])[1]/sum(as.vector(sentiment_counts)))*100
theme_sentiment_split = c(theme_sentiment_split, str_neg_percent)
mod_neg_percent = (as.vector(sentiment_counts["Moderate Negative"])[1]/sum(as.vector(sentiment_counts)))*100
theme_sentiment_split = c(theme_sentiment_split, mod_neg_percent)
neu_percent = (as.vector(sentiment_counts["Neutral"])[1]/sum(as.vector(sentiment_counts)))*100
theme_sentiment_split = c(theme_sentiment_split, neu_percent)
mod_pos_percent = (as.vector(sentiment_counts["Moderate Positive"])[1]/sum(as.vector(sentiment_counts)))*100
theme_sentiment_split = c(theme_sentiment_split, mod_pos_percent)
str_pos_percent = (as.vector(sentiment_counts["Strong Positive"])[1]/sum(as.vector(sentiment_counts)))*100
theme_sentiment_split = c(theme_sentiment_split, str_pos_percent)
}
```
```

Models developed in **R** by CX Analysts that are found valuable by business users are candidates for addition to CloudCherry’s statistical framework for general availability over standard API.

## Custom Reports

Leverage our ready integration with R Studio to create entirely custom reports for your organization with full flexibility for branding, layouts, visualization and export preferences.

These reports can be automatically updated by synchronization with CloudCherry over secure APIs.