Become a Pro-User in less than 10 minutes, and start answering questions you didn't know you had.
Let's start by getting your data into Graphext. You can directly upload a CSV, XLS, JSON, SAV, GML, Arrow, or ZIP file, connect your database using one of our 15 integrations, including Big Query, Snowflake, Amazon S3, and more... (you can connect to 10s of millions of rows of data via our integrations), or just select one of our data examples.
If you want to follow along with the examples below, we will be using the classic Titanic Survivors Dataset.
After creating your project, the first step is to work on the data section. Structuring datasets in a clear and organized manner makes it easier for you and others to manage data in Graphext. You can reorder your data and cast your variables in the data section.
The Cross Variable section in Graphext enables users to analyze relationships between different variables in their dataset by creating visualizations that combine them. This helps reveal patterns and correlations between data attributes, aiding accurate predictions and decision-making. By selecting variables to cross-filter, you can visualize their interconnectedness using various graphs and charts.
This helps uncover potential opportunities and develop targeted strategies based on insights.Once filtered, the representations in blue of the new filtered distributions indicate how the survivors differ from the rest of the population.
Graphext is great for exploratory data analysis, with a wide range of graphs that are easily accessible. The Plot section offers multiple options, including bar charts, box plots, area charts, line charts, and heat maps. Additionally, you can explore other sections like Compare and Correlations to analyze your data further.
Graphext's Wizard section provides a straightforward way for users to create models for different data types. You can create various models depending on the data type you are working with. For this example.; we will run a prediction model (although Graphext supports other types, including text, e-commerce, and location analysis).
When creating a model, it's important to choose your target variable carefully and only select the variables that you want to influence your model.
Certain types of variables will not be available for analysis. To avoid errors, ensure you have not selected problematic types or variables.
The Graph represents your data as a network by clustering similar data points into groups. Each row in your data is represented as a node on the graph that is connected to other data points by links. These links are calculated based on similarities between the data points you want to analyze, making discovering communities or patterns in your data simple.
To see which variables influence each cluster more, simply select it, and a description of those variables will appear.
You can customize the look of your graph using the graph manager. Here are some ways to do that:
Overall, the Graph tool makes it easy to explore and understand the relationships between data points in your data. By using clustering and visualizations, you can quickly identify patterns and insights that might not be apparent otherwise.
Whenever we discover an insight in our data, we can capture it by saving the current visual representation of the data to an insight card. For example, we found that females were more likely to survive the Titanic disaster. You can export the image or save it as an insight to share this insight. It's important to note that you can filter and segment your data before saving an insight.
Once saved, the insight appears in the insights panel, a space for capturing your discoveries and allowing you to present your findings. You can customize cards with charts, text, and statistics. Clicking the play icon at the bottom of insight lets you return to that point in your analysis.
Applying the model: After being satisfied with its performance, you can use it to make predictions on new, unseen data. Simply upload a dataset with the same format and select "Predict using a pre-trained model." This allows you to apply your prediction model to new daily business data, transferring learnings and increasing reproducibility in your team.
We know that data isn't always clean and simple.
Have a look through these topics if you can't see what you are looking for.