Q: Can you recommend a completely zero-based Python book? I don’t have the data base.
A: Concise Python Tutorial
I have several aliyun lucky coupons to share with you. There will be special surprises when you buy or upgrade aliyun corresponding products with the coupons! Take your lucky tickets for all the products you want to buy! Do it. We’re about to run out of them.
Q: Student in college, trying to figure out his future career.
A: I see positions related to data science, such as: Data analysis, data mining, machine learning, natural language processing, computer vision, deep learning engineer, etc., as well as recommendation algorithm and search algorithm.
I think our training objectives are closer to the requirements of data mining and machine learning. Of course, positions like data analysis, natural language processing, computer vision, deep learning, recommendation algorithms, etc., are more specialized or vertical.
I’ve always felt that machine learning is not yet a universal technology, and probably won’t be for a long time. So rather than focusing on specific technical directions, it’s more important to pick a field that interests you.
I just checked and there are more and more kinds of AI-related jobs now.
**Q: Many people in the group have foundation, what about us who have no foundation? I just didn’t know where to start. ** A: The shortcut is to take notes for every class! The purpose of taking notes is to cover all the information in the video that you find valuable! In other words, watch the video once and never watch it again. Just watch your notes!
The other thing is to spend several hours studying every day. This is very important, because you need to exercise your feeling and familiarity with data every day.
**Q: What if you are not good at math and programming? Is it still possible to study data science? ** A: Math and programming are like drilling A tunnel through mountains. Sometimes you feel like you can’t break A rock. Don’t give up. Just because you can’t work here doesn’t mean you can’t work elsewhere. Put the problem aside and follow the teacher to continue walking. It might be easier to come back to that question in a few days. Take your time. Learning takes a long time.
Just because you don’t have calculus doesn’t mean you can’t do data analysis. The influence of calculus on data analysis algorithm is regression, and other factors such as Bayes and decision tree are irrelevant. And if it’s just data analysis, it’s more important that you write Python. Python integrates all the existing model algorithms, but for the user, it’s just a matter of adjusting a few parameters, not designing the core code of the algorithm. If you want to be a data scientist, you need to understand the algorithms themselves. If you’re just playing games, the algorithm isn’t that important.
If you want to quickly catch up on math, I can recommend a good basic math video.
Q: Do you have Python installed on MAC?
A: The Mac comes with Python. To do this, enter Terminal.
Go to the application below, double-click to open it, and type Python. The Python command line environment comes with Version 2.7 of Python.
Google this link:Yanghao.org/blog/archiv…
Zhuanlan.zhihu.com/p/25198543 this…
Q: Both examples are saved in.py mode, but my data is saved in.ipny mode. Are these two because of different editors?
By default, jupyter Notebook files will be sent to Newscity in. Ipny format. You can also save them to. Python IDE, Pycharm and other common ides support. Py files.
Q: It seems that XGBOOST is using a high frequency, so the generalization performance should be good.
A: Known as big kill used to know, but the accuracy is too high, it is possible to fit, so cross verification to see.
**Q: Ta, I have a question for you. I have read Data Analysis using Python, Machine Learning, SQL and our preview materials at the same time in the past two days, and my head is almost exploding. Can you give me some advice on how to study now? ** A: I’m going to look at scikit-learn. I’m going to watch it for pandas. Master skills first, and then make up theoretical knowledge on the basis of practicing skills. As for the two books you just said, they can be used as auxiliary learning later.
For example, if you use sciKit-learn, it will be very useful to learn python and pandas.
Q: ST,data,ss2 = Summary_table (res,alpha=0.5) What are the three numbers to the left of “=”?
A: From the naming of the variables, the values returned are the standard deviation, the data, and the difference between the fitting variables. St is the entire table returned, that is, the returned values, corresponding to the first three. Data is the number for SS2. For example, the third column is the predicted result.
Q: What are the common methods of outlier detection?
A: UpperLimit=Q3+ 1.5iQR =75% quantile + (75% quantile -25% quantile)
Q: Can jupyter cancel the command? That is, after executing a piece of code, you modify the original content, and you want to undo the execution of the code and restore the original content.
A: Jupyter is A notebook, so things that are not needed can be deleted and rewritten directly, just like Word. If you are referring to how to undo the running content, press CTRL + C. It is equivalent to the content in your submission has been processed by the error code, and the processing result has replaced the original value. Now you want to do it again, so you can only load the data in your submission again, so you should be a little careful about the substitution operation when processing. Make sure your important data is backed up, such as adding new data to a column rather than replacing it.
Q: What kind of laptop is good for machine learning? Like go play Kaggle after graduation or continue to strengthen your hand.
A: Normal will do. The bottleneck of ordinary machine learning is feature engineering, not running algorithms. Notebook words brain recommended MBP, budget can consider Windows notebook, many options, Dell, XPS13, XPS15… A macbook or XPS is generally recommended.
Deep learning, CPU can run simple, complex generally not run on the laptop… Will explode. Deep learning is either desktop or remote.
Q: Data [[‘ dishonour ‘]] and data[‘ dishonour ‘]. Data is a dataframe.Disbersed is a key. Data [‘ dishonour ‘]. Apply (lambda x: sum(pd.isnull(x))) TypeError: ‘bool’ object is not iterable. Data [[‘ dishonour ‘]]. Apply (lambda x: sum(pd.isnull(x))).
A: In the Dataframe, the effect is called the index. To use a [] means to fetch a column of data. When nesting occurs, it is common to fetch multiple columns of data simultaneously.
Q: In the preview, the lecture notes are not difficult. But in operation, it’s hard to remember the related functions, so any good suggestions from your TA?
A: It is not practical to remember all the related functions, but to master the basic call method, many functions are similar, including the number of arguments and so on. Also, get in the habit of checking manuals. Many functions remember the function name, go to the manual to check the specific need to feed what parameters, some commonly used functions, with much, will naturally remember.
Q: What does Cumsum mean? Cumprod?
A: Sample accumulation, cumulative product. Suppose you have a column of data: 1,2,3,4,5. Cumsum is the sum of 1,3,6,10,15. Cumprod is the same as cumsum.
Q: In the visual preview material. Y = stats.norm.pdf(x,0,1), PDF. Parentheses. What does 0, 1 stand for? What does PDF stand for? Short for what?
A: Sample mean and standard deviation, normal Distribution sample, in PDF terms, only Probability density function, Probability Distribution function.
Q: Why do we use np.sqrt in Python for example? Plot using plt.plot; Always add a prefix. Why not use SQRT, plot directly like Matlab?
A: Most mathematical operations in Python use functions encapsulated in the Numpy module, as well as other graphing functions, which are defined in the corresponding module. Python references are module names or aliases. In addition, built-in functions can be called directly by function names, such as sorted
Ta share: The Art of Asking Questions
Address: wiki. Fast. Ai/index. PHP/H…
Questions should include:
- What did you do to solve the problem? What effects do you expect these actions to have and what are the actual effects?
- What do you guess is the root cause of the problem and what you think might be needed to fix it?
- Please tell us your basic computer environment: e.g. what system? In the cloud? Library version, etc
- Note if your steps are different from the notebook template in this course
- Error messages you receive (important!)
- If you can, please take screenshots!
- Please point out if the results are inconsistent with those returned in the lesson
- What did you do with some of the problems you’ve solved? If you reboot your computer? Restart Kernal?
- If you find some possible solutions, you can post them, such as from CSDN, simple book, etc
The above is a “question instructions” provided by data scientist Daniel, and you should try to refer to them. The clearer you are, the more others will understand. Otherwise only people who have faced the exact same problem know what you’re talking about… Stackoverflow.com has a similar question, but I didn’t bother to find one.
Teaching assistant sharing: “up and down” skills
To put it bluntly, it is their ability to find answers, and to put it bluntly, it is to use Baidu and Google search.
The first thing to remember is this truth: “In theory, almost any question I have, I can find an answer on the Internet.”
So when using Baidu, search for: keywords (such as python print difference), or directly to the error message (generally contains error) posted on the.
However, Google always has better and more relevant results in English, and you will find that typing relevant English keywords will bring more relevant answers and comments.
Finally, one more sentence: WHEN I often encounter some error information, I do not know what the meaning is or how to solve it, Baidu did not find the answer, but Google search, incredibly easy to find a lot of relevant information, and even some people asked the same question as me.
Q: I feel that I watched more videos online than I did before. How can I improve further?
A: There’s just one secret to learning Python — just use it, test it, and don’t be afraid. Study based on tests and complete assignments.
Originally I was also a mouth player, in order to complete a competition, three days to write thousands of lines of code, also read thousands of lines of other people’s code, and then I also became a revolutionary fighter. Pick up the keyboard is dry! Learn with a purpose, not lost in the ocean of knowledge. Because some things may never be used.
In addition, my personal experience, if you want to improve, you must stick together, and excellent people stick together!
This is what sets Xinniu’s course apart from other training institutions. It has teaching assistants and a cloud training platform. Hopefully, this class will take you through the whole data science work and give you a little bit of a feel for it.
Learn more about click.aliyun.com/m/34635/