You'll need to upload a .twbx file in order to get some help.
You've uploaded a .twb file. The .twbx file is a packaged workbook which contains the data. You've only uploaded the XML schema. Thx, Don
it said my file as .twbx is too large, i cannot upload it.
By looking at the graph , what do you say about relationship between sale vs quantity? I am really new to this data analysis. Thanks.
Try adding a data source filter and then extracting a smaller subset of the data to be shared.
1 of 1 people found this helpful
Visakha - Without getting into much details I'll keep this simple. The example I've used compares the correlation between Sales Qty and Sales Amount for all the Products across all 3 Product Categories.
When performing Linear Regression as this, it is important to note 3 things -
1) Slope (y=mx+b)
y is dependent variable; x is explanatory, m is slope of the line, and b is Y intercept.
2) R-Squared Value (tells me how confident my model fits, the higher the better but not always),
In my model, the R-Squared values for all three categories is pretty less (worse), which means there isn't much correlation except the fact that although my quantity is less, due to the high price, my sales amount is much higher.
Refer to this article for more on this topic,
I really appreciate your explanation. Since it is first time using this software, I really have hard time trying to understand the P-Value.
Trend Lines Model
A linear trend model is computed for sum of Sales Dollars given sum of Sales Quantity. The model may be significant at p <= 0.05.
( Sales Quantity + intercept )
Number of modeled observations:
Number of filtered observations:
Model degrees of freedom:
Residual degrees of freedom (DF):
SSE (sum squared error):
MSE (mean squared error):
I don't know how to do that. I have never used this software before.
Visakha- I will try to put here what I remember from my Stats class. P-Value is probability value in Statistics. So as an Analyst before diving into the data you'd have a hypothesis and your goal of analysis would be to prove or disprove the hypothesis in either case you back it up with some results. That's when measures like P-Value or R-Squared value come in handy.
In this example, you'd think that there is a strong correlation between Sales Quantity and Sales Amount. This is your hypothesis and in statistical terms it is called as "Null Hypothesis". Then there's an "Alternative Hypothesis" that states quite opposite. Your aim when constructing the graph and adding a trend line and thereafter analyzing the test results is to either reject or support the "Null Hypothesis". How would you do that is, using P-Value. A p-value is always between 0 and 1.
1) If P-value < 0.05, then the analysis returns a strong evidence against your claim that Qty and Sales Amount are correlated. Hence you'd have to reject "Null Hypothesis"
2) If P-Value > 0.05, then you cannot reject "Null Hypothesis" (opposite to 1)
If it is somewhere in the middle you'll have to be cautious.
Coming back to the example, Tableau returned a p-value ( < 0.0001) which is far less than 0.05. This means there is very very little evidence that Qty and Amounts are correlated.
Similarly, R-Squared value gives you the confidence level of supporting the claim. In our example again, this value is very low.
Hope this explanation helps.
If you have further questions, please attach a packaged workbook as Patrick suggested. And how to apply data source filters is to go to the data pane,
Click on filters on top right.
Click on the Add button, and from the pop-up limit your data using dimension filters (eg- date)