*Python has seen, over the last few years, a meteoric rise among data scientists, overtaking longtime rival R as the overall preferred language for data science, as demonstrated by a quick search for the terms `Python Data Science`, `Python Machine Learning`, `R Data Science`, and `R Machine Learning` on Google Trends:
***Numerical type**: The user can select a mathematical comparator and enter a value for the operand. For convenience, the UI will also show statistical values related to the chosen column, which can be used when picking the operand value:
1.The PixieApp is published into the PixieGateway server directly from the Notebook and a URL is generated. Behind the scene, PixieGateway allocates a Jupyter Kernel to run the PixieApp. Based on configuration, the PixieApp could share the kernel instance with other apps or have a dedicated kernel based on needs. The PixieGateway middleware can scale horizontally by managing the lifecycle of multiple kernels instances, which themselves can either be local to the server or remote on a cluster.
***Method arguments**: This is useful when the route arguments use the `*` value and you want to have access to that value at runtime. In this case, you can add arguments to the method itself using the same name as the one defined in the route arguments and the PixieApp framework will automatically pass the correct value.
2.Download and install the Kubernetes CLI ( [https://kubernetes.io/docs/tasks/tools/install-kubectl](https://kubernetes.io/docs/tasks/tools/install-kubectl) ) and the IBM Cloud CLI ( [https://console.bluemix.net/docs/cli/reference/bluemix_cli/get_started.html#getting-started](https://console.bluemix.net/docs/cli/reference/bluemix_cli/get_started.html#getting-started) ).
***Storage connector**: Configure a persistent storage for various resources, such as charts, and Notebooks. By default, PixieGateway uses the local filesystem; for example, it will store the published Notebooks under the `~/pixiedust/gateway` directory. Using the local filesystem is probably fine for a local test environment, but when using a Kubernetes install, you will need to explicitly use persistent volumes ([https://kubernetes.io/docs/concepts/storage/persistent-volumes](https://kubernetes.io/docs/concepts/storage/persistent-volumes)), which can be difficult to use. If no persistence strategy is put in place, the persisted files will be deleted when the container is restarted and all your published chart and PixieApps will disappear. PixieGateway provides another option, which is to configure a storage connector that lets you persist the data using the mechanism and backend of your choice.
***Run code**: This is code that will be run in its own instance for every user session. The run code is typically extracted from the cell that contains the PixieApp class declaration. The publisher automatically discovers this cell by doing a static analysis of the Python code and specifically looking for the following two criteria, which must both be met:
***Security** : Configure the PixieApp security when accessed through the web:
***安全**：通过网络访问时配置PixieApp的安全性：
* **No security**
* **Token** : A security token must be added as a query parameter to the URL, for example, `http://localhost:8899/GitHubTracking?token=941b3990d5c0464586d67e48705b9deb` .
*Don't forget to add the code that runs the PixieApp in the same cell. If not, you'll end up with a blank page when running it on the web. As a good practice, it is recommended to assign the PixieApp instance into its own variable. For example, do this:
*The `pd_app` attribute provides a cleaner separation between the components and more flexibility than widgets. You could, for example, have a button that dynamically invokes multiple PixieApps based on some user selection.
*A simple PixieApp that demonstrate how to create streaming visualizations from randomly generated data: [https://github.com/pixiedust/pixiedust/blob/master/notebook/pixieapp-streaming/PixieApp%20Streaming-Random.ipynb](https://github.com/pixiedust/pixiedust/blob/master/notebook/pixieapp-streaming/PixieApp%20Streaming-Random.ipynb)
***Programmatic**: In some cases, you may want to directly trigger an event via JavaScript. In this case, you can use the `sendEvent(payload, divId)` method of the `pixiedust` global object. The `divId` is an optional argument that specifies the origin of the event. If the `divId` argument is omitted, then it defaults to the `divId` of the element that is currently sending the event. As a result, you should always use `pixiedust.sendEvent` without a `divId` from a JavaScript handler of a user event such as click, and hover.
2.Use the `@PixiedustRenderer` decorator to register the `rendererId` (which must be unique across all renderers) and the type of chart being rendered.
@@ -811,7 +811,7 @@ class SimpleDisplayWithRenderer(BaseChartDisplay):
pdb.run("<insert a valid python statement here>")
```
*By setting an explicit breakpoint in the code with the `set_trace()` method:
*通过使用`set_trace()`方法在代码中设置一个明确的断点：
```py
import pdb
...
...
@@ -838,7 +838,7 @@ class SimpleDisplayWithRenderer(BaseChartDisplay):
魔术命令有两种类型：
***Line magics**: The syntax is `%<magic_command_name> [optional arguments]`for example, `%matplotlib inline`, which configures Matplotlib to output the charts inline in the Notebook output cell.
***Cell magics**: The syntax is `%%<magic_command_name> [optional arguments]`. For example, we call the HTML cell magic to display HTML on the output cell:
2.Create the bottleneck values (caching them as appropriate) that normalize the image by resizing and rescaling it. This is done in the following code:
