Project portfolio C.Jacobs

Demo-graphic

The first project i completed is called Demo-Graphic. This is a dashboard programmed in R that uses demographic data from Kaggle on liver disease. Within the dashboard it is possible to interact with the database, analyze data, filter data, download reports and manage data and users.

UPLOAD DATA

Demo-graphic includes an upload panel that adjusts the column structure and ensures data integrity during the upload process to the database.

SQLite DATABASE

Experiment data and user data are stored in an SQLite database, with user information encrypted for security.

DATA/USER MANAGEMENT

User and experiment data can be managed through an admin account. Faulty entries, which are uniquely identifiable, can be removed with administrative privileges.

FILTER DATA

For analysis, it is important to filter relevant data, a feature available within the interface

ANALYSIS

Various graphs and images can be generated for analysis, allowing the investigation of frequency distributions and correlations.

REPORTING

A report can be downloaded containing all the filtered data and the generated graphs.


Login

The login screen allows users to enter their credentials, with an option to contact the application administrator if needed.

Upload Page

The upload page features a tab for registering experiments, with this information stored in a separate table in the database. When uploading data, an experimental ID is required. The program verifies that the values in the tables meet specific criteria; otherwise, an error is displayed. The username of the logged-in user is automatically associated with the experiment. Columns are automatically matched if a similar name is found in the database, encouraging users to streamline their workload by using standardized templates.

* Unique Data Entries *

To ensure that data entries are uniquely identifiable, the program adds a key along with a timestamp that includes the day, month, year, and time. This feature allows the application administrator to delete incorrect entries when necessary.

Filtration

Data of interest can be selected within the filtration tab. Data is loaded from the SQL database and manipulated within R. While this process is efficient for the current data size, it may become less efficient as the data grows, potentially requiring more selective queries. Users can only access experiments in which they were involved; data from other experiments remains private unless the user has worked on the same experiment. The filter supports the filtration of both discrete and continuous variables.

Data visualization and Analysis

The application allows for data visualization. The data displayed and analyzed corresponds to the selections made in the filter tab.

* Explore Frequencies *

In the analysis tab, you can explore the frequencies of different continuous variables and adjust the bin width as needed. These continuous variables can then be categorized to examine occurrences within each bin. Additionally, the frequencies of categories can be analyzed in the categorical tab, which displays the occurrence of each category as a count (N) and allows for the division of these categories into subcategories.

* Explore Distributions *

To explore the distribution of occurrences, you can create both violin plots and box plots. Box plots offer a clear view of the spread and skewness of the data. Violin plots, on the other hand, combine the features of box plots with density plots, providing a more detailed representation of the data distribution, including the probability density at various values.

* Violin Plot *

* Box Plot *

* Explore Correlations *

Exploring correlations is an effective way to identify variables that may contribute to a particular outcome, such as diabetes. To facilitate this exploration, a correlation matrix has been added. You can also select the statistical test used to generate the matrix.

Reporting

A record of the filtered data and associated graphs can be created by clicking the download button in the report tab.

* Download Report *

The report is saved in a compressed folder named plots_and_data_YYYY_MM_DD, as displayed in the download folder.

* Excel File *

The compressed folder includes an Excel file that records the data used for analysis, allowing users to perform custom investigations.

* PDF File *

The PDF file within the compressed folder includes all the graphs created in the dashboard. The file is automatically adjusted to reflect the specific graphs generated.

Data Management

Mistakes can occur, and uploaded data may contain incorrect entries. To address this, a data management tab has been created for profiles with administrative rights, allowing for the removal of erroneous entries.

* Delete Data *

Uploaded records receive a date-time stamp, making them uniquely identifiable in the database. To prevent the manipulation of results by removing specific entries, only entire file uploads can be deleted. To remove a file, the user must first select the correct data and then validate it by copying the entries into a textbox. The user can proceed with the deletion only when the selected variables match those in the textbox.

<User Management>

To facilitate user management, the R Shiny package shiny-manager was utilized. This package enables administrators to manage users, set passwords, and monitor activity. User information is securely stored in a local encrypted SQL database, ensuring that all data is kept safe.

* Monitoring Users *

Application usage can be monitored through the interface. Logs of this activity can be downloaded for further analysis and are also visualized within the dashboard. These logs display details such as who logged in, when they logged in, their access rights, and whether the login was successful. The first image below shows the interface in the dashboard, while the second image displays the downloadable data.