Glassdoor Employee Reviews

This dataset was compiled by a Kaggle.com user in Decmeber 2018. This user scrapped employer reviews for some of the top tech companies from Glassdoor.com and aggregated them into one dataset. The dataset has 67529 rows and 17 columns. Each row represents one employees review. The columns are as follows:

## Index(['Unnamed: 0', 'company', 'location', 'dates', 'job-title', 'summary',
##        'pros', 'cons', 'advice-to-mgmt', 'overall-ratings',
##        'work-balance-stars', 'culture-values-stars',
##        'carrer-opportunities-stars', 'comp-benefit-stars',
##        'senior-mangemnet-stars', 'helpful-count', 'link'],
##       dtype='object')

Graph 1 - Average User Ratings by Category and Company

This first graph displays the overall rating of each company captured in the dataset and averages each of their user reviews. User reviews rate a company on 4 factors, work life balance, culture, opportunities, comp & benefits, and provide an overall rating as well. This graph ranks companies by their overall average rating across all reviews, but also breaks out their average review score in each subcategories to provide a picture of where companies excel and where they could possibly improve their employee experience.

## ([<matplotlib.axis.XTick object at 0x000000002858B908>, <matplotlib.axis.XTick object at 0x000000002858B320>, <matplotlib.axis.XTick object at 0x0000000028038FD0>, <matplotlib.axis.XTick object at 0x000000002861F240>, <matplotlib.axis.XTick object at 0x000000002861F6D8>, <matplotlib.axis.XTick object at 0x000000002861FBA8>], <a list of 6 Text xticklabel objects>)

Graph 2 - Average User Ratings by Category and Company

This chart breaks out the number of reivews each company recieved by the percentage of total reviews. This tells us which company recieves the most reviews and which company has recieved relatively little. The goal here was to identify companies that had a large pool review so that later we can break out some of the top phrases their employees say about them.

## ([<matplotlib.patches.Wedge object at 0x00000000324AF898>, <matplotlib.patches.Wedge object at 0x00000000324AFF98>, <matplotlib.patches.Wedge object at 0x0000000029FAD6D8>, <matplotlib.patches.Wedge object at 0x0000000029FADDD8>, <matplotlib.patches.Wedge object at 0x0000000029FB5518>, <matplotlib.patches.Wedge object at 0x0000000029FB5C18>], [Text(1.218072800222186, 0.45420111554121345, 'amazon'), Text(-0.7533978162891347, 1.0594299082104315, 'apple'), Text(-1.2507266733748335, 0.35451768433848657, 'facebook'), Text(-1.2830787018057983, -0.20906708246959232, 'google'), Text(-0.2727468872924693, -1.2710661412657758, 'microsoft'), Text(0.7975127951168561, -1.026632038086139, 'netflix')], [Text(1.0775259386580875, 0.40179329451722723, '39.1%'), Text(-0.6664672990250038, 0.9371879957246123, '19.2%'), Text(-1.1064120572161986, 0.3136117976840458, '2.4%'), Text(-1.1350311592897444, -0.18494395756925472, '11.6%'), Text(-0.24127609260487667, -1.124404663427417, '26.6%'), Text(0.7054920879879881, -0.9081744952300459, '1.2%')])

Graph 3 - Reviews Over Time

I wanted to explore whether the average review had gotten better or worse for each company over time. This chart looks at the average overall score for each company by year and helps trend that rating Some companies have improved their overall employee experience while others have stayed stagnant. Data like this coupled with Graph one can help employers gain insight into what they’ve been doing right and whether their efforts have had any long term impacts to how their employees perceive their organizations

## <matplotlib.axes._subplots.AxesSubplot object at 0x0000000025868358>
## <matplotlib.axes._subplots.AxesSubplot object at 0x0000000025868358>
## <matplotlib.axes._subplots.AxesSubplot object at 0x0000000025868358>
## <matplotlib.axes._subplots.AxesSubplot object at 0x0000000025868358>
## <matplotlib.axes._subplots.AxesSubplot object at 0x0000000025868358>

Graph 4&5 - Most Frequent Pros and Cons

Finally, I wanted look at the most reviewed companies pro’s and con’s. Each reviewer not only wrote a summary review, but also wrote a pro and a con for that employer. One of the easiest ways to get an accessible look into a mass of text is to use a word cloud. I put together two word clouds, one for Pros and one for Cons based on the most frequent words in Amazon’s reviews. The results clearly show that Amazon employees like their benefits, but don’t like their managers. An insight like this can help lead deeper dives into uncovering employee concerns.

## (-0.5, 399.5, 199.5, -0.5)

## (-0.5, 399.5, 199.5, -0.5)

Source: https://www.kaggle.com/petersunga/google-amazon-facebook-employee-reviews