Читать книгу QlikView Your Business - Troyansky Oleg - Страница 8

For readers unfamiliar with QlikView or Qlik Sense, this section describes the core elements of the platform, and explains how it differs from traditional BI.

In-Memory Storage Means No Need for Pre-Calculated Cubes

Unlike OLAP systems, QlikView uses RAM as the physical storage medium for data. Since computers can access memory hundreds of times faster than disk, calculations and aggregations can be performed on the fly, with astounding speed. Thus, the limitations of building pre-aggregated cubes are gone! Figure 2-2 shows a data ecosystem with the addition of QlikView. Notice that QlikView can extract from multiple sources and does not require any pre-aggregated cubes.

Figure 2-2: QlikView in the data ecosystem

Using QlikView, transactional data can be loaded into RAM and then summarized at runtime, at the user’s request. If you’re used to the terminology of traditional BI, you can think of QlikView as creating “cubes on-demand,” from RAM. Transparent to the user, the aggregations occur seamlessly in the front-end, with each chart essentially creating its own cube. The benefit of loading the granular detail is two-fold: the data can be aggregated up to any level, and the user can drill down to view the details.

For the small or medium-sized organization, QlikView may replace the need for OLAP or other reporting tools. For the enterprise, QlikView is often added as a data discovery/analytics platform that works alongside OLAP systems – particularly if the organization still requires paper-based reporting.

An Interactive User Experience

A user accustomed to the traditional BI report interface knows that you need a game plan going in, before actually seeing any data. Typically, the user must select a specific report and provide the required parameters or filters before the report is run. QlikView completely rejects this approach, and instead presents the user with all of the available data, immediately accessible in the interface.

When a user opens a QlikView application, data is visible right away, without specifying any parameters. The user interacts with the interface to step through the data in an exploratory way, to zero in on specific results. Figure 2-3 shows a basic example of QlikView application containing sales data for an apparel company. You’ll use this data set, which is available from the book’s download site, throughout the book.

Downloading the Electronic Materials for This Book

If you haven’t done so yet, please download the electronic materials provided for this book. You can find the detailed instructions at the end of this book’s Introduction.

This is a screenshot of a typical QlikView application. Using a tabbed sheet layout, developers place objects on the sheet to allow for searching and selecting data and visualizing measures.

Figure 2-3: Example QlikView 11 application

In this app, a few filter objects called list boxes are shown across the top (Year, Quarter, Month) and down the left pane (Channel, Product, Season). Three visualization charts are shown: a pie chart, bar chart, and straight (non-pivot) table. The data in the charts reflect the entire data set, with no filters applied. In QlikView parlance, filters are called selections. The current state of the selections can be tracked in the Current Selections box, shown in the upper left.

Using a familiar tabbed sheet layout, this simple QlikView application invites the user to make selections to explore the data and click on the tabs to explore the layout. By default, selections made on one tab are persistent throughout the entire application (this behavior can be changed by the developer, depending on requirements). In Figure 2-4, the app is shown with selections applied for Channel and Season.

Figure 2-4: Filters applied in a QlikView 11 application

As soon as the user makes selections, the data in the charts dynamically update. No need to press Go, Generate, or Apply – results are rendered immediately. All visual objects in a QlikView app can be interactive in some way. Most obviously, the list boxes allow clicking or searching for attribute values. In addition, the user can click or lasso the slices of a pie chart, or the bars in a bar chart, to make selections within the visualization itself. The table in the bottom-right of Figure 2-4 is also selectable and sortable.

With an attractive and interactive interface, QlikView apps encourage users to ask questions of the data, which may encourage asking questions of each other, which may result in collaboration, which may then lead to true business insight.

Associative Logic Powers Data Discovery

Perhaps the most effective driver of data discovery in QlikView is its patented associative query logic. Without going into the details of how it works, let’s look at what it delivers.

Selected, Associated, and Non-Associated Values

Perhaps the most obvious feature of QlikView’s associative logic is the ability to visually see how other pieces of data are associated with your selections. The feature that truly differentiates QlikView is the ability to see the data that is not associated.

Figure 2-5 shows list boxes with two explicit selections applied.

Figure 2-5: Green, white, and gray

Selections are made in the Product Group and Warehouse list boxes for Casual and Memphis, respectively. What can you learn from this simple selection?

Based on QlikView’s display defaults:

● Selected values are highlighted in green

● Associated values are highlighted in white

● Excluded values are highlighted in gray

From this, you can infer the following:

● There are no products from the Spring collection in the results

● There are no customers from ID, HI, NE, NM, RI, or VT

● There are no products from the Q-Tee Golf brand in the results

Using only list boxes, QlikView can visually communicate meaningful associations within the data. Seeing which data is associated (and which is not) can confirm a hunch or prompt the user to look under previously un-turned stones. A question of, “Why are there no Q-Tee Golf products shipping out of Memphis?” may lead the user to one of these conclusions:

● The data is wrong and needs to be fixed or cleansed

● There is a business problem that needs to be solved

● There is a valid reason for the results

All of these are valuable outcomes! The power to show you what is related and what is not related is a key feature of Qlik’s patented associative logic.

Exercise 2.1: Experience Green, White, and Gray in QlikView 11

1. Open QlikView.

2. To open an existing document, use the menu command File ⇒ Open. Navigate to the folder containing the electronic materials for this book, subfolder \Apps, and open the document Example Sales Analysis.qvw.

3. From the Sales tab, select three products in the Product list box – Baby Jacket L Black, Blue, and Green, as shown in Figure 2-6.

Figure 2-6: Product selections

a. 3.1. Click on the first product in the list and hold down the left button while dragging down. Let go of the mouse button when you’re hovered over the third product in the list. The Product list box should look like the one in Figure 2-6, with the three products highlighted in green.

b. Notice the values associated in other fields (in white) – as well as the values not associated with your selections (in gray).

4. Explore the data by making other selections and viewing the charts on the Trends tab. Notice that selections are persistent as you navigate tabs.

5. Clear selections by clicking on the Eraser icon in the caption of each list box, or within the Current Selections box.

In addition to the common green, white, and gray selection states, Qlik Sense offers an easier way to handle the lesser-known states of Alternate and Select Excluded. These states are described in Chapter 16.

Direct and Associative Searches

QlikView’s associative logic offers the ability to conduct direct and indirect (associative) searches either within a single field or the entire data set. To demonstrate, let’s take a look at the bread-and-butter visualization object in any QlikView application – the list box. List boxes are used to display the values found in a single field in the data model. Values can be selected by directly clicking on them, or by searching. The direct and indirect search capability is illustrated in Figure 2-7.

Let’s assume that you want to search for a product in the Product list box. One way to initiate the search is to click on the magnifying glass in the caption of the Product list box and type in a search phrase. In this example, the phrase “jacket” is typed in the search box. The direct search results appear in yellow highlight in the Product list box. Associated results appear when you click on the chevron (>>). Notice that the phrase “jacket” was found in two other fields, Style and Style Short Name. You can either click on the exact results in the Product list box or select from the associated results to temporarily limit the products in the Product field.

Figure 2-7: Direct and associative search results

Exercise 2.2 describes the basics of the text search capability in QlikView.

Exercise 2.2: Search and Associative Search in QlikView 11

1. Open QlikView.

2. To open an existing document, use the menu command File ⇒ Open. Navigate to the folder containing the electronic materials for this book, subfolder \Apps, and open the document called Example Sales Analysis.qvw.

3. Search for “jacket” in the Product field.

a. Click on the caption (title) of the Product list box and begin typing jacket (the string is not case-sensitive). You can also click on the magnifying glass in the caption area of the Product list box

b. Click or Ctrl+click on individual products or press Enter to select all of the products returned in the search. Notice that the contents of the Current Selections box display the filters that are applied.

c. Right-click on the Product list box and select Clear (or click the Eraser icon next to the magnifying glass).

4. Limit the results for “jacket” by using Associative Search.

a. Click on the caption of the Product list box and start typing jacket.

b. Click the chevron (>>) in the right corner of the search box to display associated results.

c. Click on Dressy Jacket in the associated results area (see Figure 2-8) and press Enter. Notice that this did not explicitly select Dressy Jacket in the Style field, but instead limited the results in the Product field to those that also have the Style attribute “Dressy Jacket.”

Figure 2-8: Direct and associated search results

5. Based on QlikView’s green-white-gray display rules, what can you learn about the Season availability of the Dressy Jacket products? Which states do not have customers with Dressy Jacket sales?

While this exercise described the common text search feature, there are several other search features available, including numeric search, fuzzy search, and an advanced search dialog. For more information on these features, open Help ⇒ Contents from the menu and type search on the Index tab.

A Front End with No Queries

Using QlikView’s built-in ETL features, data from source systems are modeled, transformed, and loaded into memory. The resulting set of data, in memory, is the source of data for the front-end objects.

Qlik’s associative architecture maintains the relationships among all data points in memory, in real time. After each selection that a user makes, the associations in the data model are updated. What does this mean? It means that the front-end objects do no not require SQL-like queries to define the object. In traditional BI systems, a data query must be written for each chart to properly fetch data from the cube. In other words, the developer must define, with SQL code, how the data is related each time a chart is created. This makes it almost impossible for non-technical users to design their own applications. In QlikView, the difficult queries are written once, in the ETL layer. The resulting data set is then available to the front end with all of the associations intact. With the data loaded into memory, charts do not require supporting queries – they only need to be configured with a dimension and a measure. With minimal training, non-technical users can create their own dashboard objects without knowing how to write SQL queries.

Right-Sized Analytics

With several deployment options, QlikView offers a right-sized solution for any analytics or data discovery project. The desktop QlikView client allows you to quickly load data and create visualizations, all from your personal computer. Anyone who’s ever used Excel to extract data from an external source can easily learn to do the same in QlikView. Analysts can use QlikView to build their own applications to answer ad hoc business questions or create compelling visuals to use in presentations. Try doing that with traditional BI platforms!

On the other end of the scale, QlikView’s server platform provides for sharing QlikView applications among teams, or thousands of users within a global organization. With the option of clustering QlikView servers, users can have highly reliable access to applications in a distributed enterprise environment.