Norsk versjon:
The code here uses data that is only available to SSB employees. Confidential data is hidden from this README and from the code. The code itself was developed in the Statistics Norway (SSB) organisation repository and later a lite version was copied here to reflect my work done for SSB.
A tool for analysing trends relating to a category of responses (such as curse words) in the NØKU statistics survey.
A lighthearted analysis of specific responses/comments/feedback received for on in the OKI-Schema delivered to the respondents of the main NØKU survey.
While a little tongue in cheek, it should be appreciated that this feedback is taken very seriously and although the analysis here is done by code, that humans actually read all the feedback that is provided. This is how we are able to provide a comphresive word list to feed into the program.
There are important messages in this data. Negative feedback is valuable, no matter how it is expressed. We at SSB have heard the message loud and clear and we are making efforts to improve the experience by respondents to surveys. Current ideas being explored include the use of AI to assist responders, the use of machine learning models to populate business statistics distributions and other statistical models are being developed with the end goal being that we dont need to send surveys to as many businesses.
Running the code is pretty simple:
- Enter as many swear words, phrases, colourful comments as you can think of. Have fun, be imaginative. Then simply run the code.
Ranking the curse words responses based on industries at a 2 digit level:
Ranking the use of '!' on industries at a 2 digit level (log scale used):
Ranking of curse word and exclamation point usage on a 3 digit industry level.
Are the usage of swear words and '!' correlative? It appears to not be the case. It seems in Norway its either one or the other.
All industries over time:
Interactive plots for 2 and 3 digit level industries over time:
