|
|
| Line 1: |
Line 1: |
| − | This is the dashboard we have just built to analyze laptops' data from Amazon.
| |
| | | | |
| − | It is currently displaying a dataset that includes laptops of different brands and series. Remember that this is a dataset that we have built by scraping data from Amazon; but the module that is intended to Load a new dataset is not ready. We are currently working on it. So when this module is ready, the application is gonna be able to scrape data from Amazon, from a page like this, in real-time.
| |
| − |
| |
| − | The other page in which we are currently working on is the Sentiment analysis page. That is also a very impotant topic for the application but it't no ready yet.
| |
| − |
| |
| − | So before showing the main features of the application, I wanted to show you how we are scraping the data from Amazon.
| |
| − | So this is the script we built with a Python framework called Scrapy, I'm gonna run it, so here we are scraping the data from amazon, it is quite fast, here we are just scraping about 35 laptops
| |
| − |
| |
| − | It has created this file, let's see the file. It a JSON file with the detail of 35 computers, so we can see all the details of the laptps, the technical details and the reviews. So when this module is ready, it will run the script we have just seen and scrape the data from amazon in real-time.
| |
| − |
| |
| − | So let's talk about the home page. So the home page has been designed to allow the user to discover and visualize the data. So you are able to customize the data you want to display by selecting the brand, series, and the range of prices.
| |
| − |
| |
| − | So let's say that you want to analyze all the brands at the same time. When you select a brand it automatically selects all the series for this brand but then you can filter the brands you want. It takes a bit because is processing the data
| |
| − |
| |
| − | Or I don't know maybe we are interested in expensive laptops. So let's select computers over 1000 dolars for example. You can see that now the application only proposes computers in this range of prices, there aren't a lot actually. You can see that they are gaming laptops which are usually expensive.
| |
| − |
| |
| − | But in other cases is better to analyze only one brand. Let's, for example, analyze Acer computers.
| |
| − |
| |
| − | The firs charts that we have included is to compare average customer reviews and prices, You can see that the blue bars show the values for all items, that means for all the series of the brand,
| |
| − | but we have also included a red bar that displays the values for selected items only. For example, is you want to know the price of a specific computer, so let's select for example this one... you can see that this is a very expensive one.. 1088$ and that has actually a very good customer review score of 4.3, so it's apparently a very good computer
| |
| − |
| |
| − | The second panel we have included is a Bubble chart that shows the Average customer reviews vs. Prices.
| |
| − |
| |
| − | We have included this chart because actually one of the main faatures that can be analyzed when talking about sales, is the relationship between price and customer satisfaction. So with this kind of chart, we would try to determine a trend to establish a relationship between price and customer review.
| |
| − |
| |
| − | One nice feature of these charts is that you can select the brand that you want to visualize. If you click in one brand this is going to be excluded the brand clicked from the chart, but if you double click, only the brand clicked will be shown
| |
| − |
| |
| − | The other panel we have included in to visualize the most frequent words in customer reviews. Word clouds provide a nice visualization of the most frequent word. But if you need to be most precise, you can use the word count chart that provides the exact number of times a word has been mentioned.
| |
| − |
| |
| − | So let's for example analyse the information provided by the wordcloud. We can see that some of the most frequent words in customer reviews are:
| |
| − |
| |
| − | Words like good of grate indicate that it is a computer that users have liked, but we already knew that customers liked this computer by analyzing the average customer reviews score that is 4.3.
| |
| − |
| |
| − | But a information that we didn't know and it's provided by the wordcloud with words like Gamming or game it that this is a Gaming laptop.
| |
| − |
| |
| − | We finally found the word “Screen”. which is probably the word that provides the most important information from this word cloud. We can see that users are talking about the screen of this laptop, but we can not be sure if they are saying something good or bad about the screen. We can actually infer that is something good based on the good customer reviews score or based in the other words that are present in the word cloud that provided a positive sentiment like great and good but in the end we cannot be sure about what customers are saying about the screen. This is why there are other analyses that can bring more information, like sentiment analysis, which is the topic we are currently working on.
| |
