Overview of App
This RShiny app is a prototype for a graphical user interface for the finnsurveytext R package. Further development is planned. If you would like to provide feedback on this app, please email adeline.clarke@helsinki.fi.
The app attempts to provide a “no-code” way for a user to engage with finnsurveytext to analyse open-ended survey responses. It does not demonstrate all the functionality of finnsurveytext (for example it does not allow integration with svydesign objects) but provides a mainstream path to initial analysis for users who do not want to code in R (yet).
The app is split into 4 tabs (along the top).
Instructions (this tab)
Prepare Data
- This tab is split into 2 sections (along the left-hand side).
- The first section (Load Data) allows the user to load in their own data in a csv format or to choose to use one of two sample datasets.
- In Format Data, the user fills out information about their survey including which columns contains the ID and the responses to the open-ended question, which Finnish treebank should be used to process the data, and whether to remove stopwords from the data. Once these have been filled out, the green button should be pressed to format the data for later steps.
- Optionally, the user can add a column in their data to be used to weight different responses and/or bring in other additional columns for splitting the data in the 4th tab Compare Groups of Responses
Explore Data
- In this tab, the formatted data is used to create some tables and visualisations.
- Firstly, the Summary Tables are used to create simple summaries of responses including which proportion of survey participants provided an answer for the open-ended question, the number and types of words used, and distribution of the length of responses.
- The Wordcloud section allows the user to create a simple Wordcloud from the formatted data showing frequent words. Users can choose how many words to include in the cloud, to exclude certain word types from the cloud, and to weight words if a weight column was provided during formatting.
- The N-Grams section creates a plot of the most frequent words/phrases so can be used to better quantify the most frequent words. Again, users can possibly weight these and also exclude word-types from the plot. Also, there is the option to normalise by number of responses or total number of words.
- The final section is the Concept Network section. In this section, users can create a Concept Network plot of responses based on chosen concept words.
Compare Groups of Responses
- In this tab, we have corresponding comparison functions for those in Explore Data. You can use a column added during Data Preparation to split responses into groups and choose whether to exclude or include responses which had a null in this column.