In-depth insights with AI and machine learning data analysis tools

So now for some of the cool new AI features. I’m back here with my hotel reviews data set that I had before as well. And so I’m in the Power QueryOnline Editor on my data flow.

Now, in the data set I have, I have some reviews that people left about hotels and they’ve also uploaded some images. Now one of the things I can do to really enrich this data is that I already have to use the AI insights.

And this shows me a couple of functions that are available to me and it allows me to just out of the box, uh, use these for instance these cognitive services. So I could detect language, I could extract key phrases.

For this example, I wanna score the sentiment of the reviews. I’m just gonna select that one. And then I can put in the field that I want to apply this function to. So I wanna do this on the review text. For now, I just have to put in the language that is gonna be in, but when we go into production, that will be automatically detected.

So then I’m gonna invoke the function and while this is running, this is running just in the Power BI service and as with any other transformationI’m doing in a power query, it will add a column with the result of that function. And so you see it just takes a second and now I have the actual sentiment score of this review.

Now maybe there are some data scientists in my organization who have created this amazing model for tagging images that’s specific to my use case. Now with this new integration with Azure ML, I can actually leverage that right out of Power BI. So my data scientist has created this model in Python using the Azure ML SDK and um, they have created web service for that model.

Now all the data scientist has to do is go into the Azure portal, give me access to the model, and as soon as they’ve done that, it will show up in my functions here. So as before, I’m going to insights right over here, and then I have my Azure Machine Learning models. And as you can see, that’s where the hotel image classifier is showing up. So I select that, and it can just as easily select the column, the image column that I wanna apply it to, and I just invoke it.

Now similar to the previous steps is all recorded here, it just applies that, uh model on the data that I’ve selected. If I expand the column here, is just gonna take a second, and doing that, it will just add the image text to my data set. Let’s look at that, so you see like pool, there are some pictures about the hotel rooms, got the views, and now I can save again that data set and connect to it from desktop or just create a report on it. Now let’s see what that would look like.

So I already have a report that is about these hotel reviews and so far, I was just able to load the images with the reviews that are related, and I can see how many reviews have been left on which hotel and which island. But by adding that sentiment and adding those text I can make this analysis much more interesting. So one thing I can do is for instance look at the sentiment score, uh, by the hotel.

Just gonna drag that on here, and make it to a bar chart, and there you can see how I can filter actually my reviews by the hotels that have the highest or lowest sentiment. Now another thing I can also do is to bring in the image text.

Wanna make a new graph with the image text, add the sentiment score, and then again make it into a bar graph. So right now I can just see which images are related to high and low sentiment. So again I can just filter on that, I can say okay, people that have images, uh, with beaches generally have a high sentiment, and I can see which hotels those are for.

And then also, the lower sentiment is in this case, about ACs that are, you know, not really looking like something you wanna find on your holiday. So these AI functions allow analysts to enrich their data and to get more depth in their analysis. But we’re also doing a lot for end-users, mostly just to make their lives easier and to enable them to find insights faster.

Now, one example here I have is a tourism analysis on Hawai’i. And so you can see how many people are visiting the different islands, why they are visiting, and how much they’re spending, but there’s also a lot of information behind this that is not shown may be directly in this report. Now with quick insights, I can actually get that out. We originally introduced quick insights to get insights on your complete data set, but now we’re actually focusing it on the points that are relevant to you, so the context that you’re looking at. Let’s look at the visits by date and region.

If I select that, and I right-click, I can now use the analyze, and Power BI is already seeing what type of data and what type of graph this is to know what would be my next deep dive. So we can say I wanna explain the decrease, and it will show me what has happened and what is driving the decrease behind this chart.

So for instance here I can see that conventions, business people visiting for conventions, are, is actually going up, but the biggest decrease is in people visiting for vacation, which might make sense if you compare August with September.

Leave a Reply

Your email address will not be published. Required fields are marked *

Previous Post

Bollywood producer and stylist Rhea Kapoor weds in style

Next Post

AI-Driven Drug Discovery

Related Posts