There are two ways to work on https://www.nngroup.com/articles/journey-mapping-approaches/ : assumptive, and research-based. In an assumptive journey map, you are working with a group of internal stakeholders and building the map based on all of the assumptions that they make about the persona on his or her journey. These assumptions can (and probably should) be validated with user research later. After you create the journey map as a group, you can use dot voting to identify pain points. The phase or action that receives the most votes would be your "lowest point" in the journey. In a research-based journey map, you would interview several users who represent your target persona, and ask them specific questions about each action and point in the process. At the end of learning about their process, you'd ask them to tell you which part of it needed the most improvement. You might receive a lot of different answers, so if you have a lot of "low points" without any really standing out, you might still want your stakeholders to come back and add additional votes to help make a decision.

If I understood the question correctly:grp = df.groupby("Датчик")["Показания"] first_idxs = grp.apply(lambda x: x.index[0]) df["счетчик"] = np.where( df["Показания"] != 0, grp.apply(lambda x: (x.index.isin(first_idxs) | x.eq(0)).cumsum()), 0 ) result:In [71]: df Out[71]: Дата Датчик Показания счетчик 0 01.01.2020 1 0 0 1 02.01.2020 1 2 1 2 03.01.2020 1 3 1 3 04.01.2020 1 5 1 4 05.01.2020 1 0 0 5 01.01.2020 1 22 2 6 02.01.2020 2 3 1 7 03.01.2020 2 0 0 8 04.01.2020 2 11 2 9 05.01.2020 2 22 2

The mistake was because I forgot to put the commas at the end of each line in the area.

Try to use the Pandas by https://pandas.pydata.org/docs/reference/api/pandas.Series.mode.html :Dts[fnc].rolling(1000).apply(lambda x: x.mode()[0])

Use it. idx = pd.IndexSlice slice_ = idx[:, "x1":"x3"] df.style.apply(highlight_max, props='color:white;background-color:green;', axis=1, subset=slice_)

Just like you said,df["new"] = df["brand"].str[0] + df["brand"].str[-1] + "-" + df.index.astype(str) result:In [22]: df Out[22]: ID brand model new 0 AI-0 AUDI TT AI-0

Group the poles. ["Код", "Название"] And make a pole. "Сумма":In [106]: df.groupby(["Код", "Название"])["Сумма"].sum().reset_index() Out[106]: Код Название Сумма 0 03.2.01.01-0001 Цемент 420 1 14.5.06.03-0002 Паста 187

Use the method. https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.dropna.html :res = df.dropna(thresh=int(len(df) * .8), axis=1) result:In [55]: res Out[55]: Col1 Col3 Col4 Col6 0 7 9.0 7 4 1 6 3.0 5 10 2 6 4.0 3 2 3 8 9.0 6 8 4 6 6.0 8 5 5 10 5.0 2 5 6 4 NaN 4 7 7 4 NaN 4 10 8 10 4.0 3 10 9 5 4.0 4 7 NOTE: You' df.dropna() removed only NaNnot gaps. If you have gaps in your column, this column will need to be converted first to the number of data:df["col_name"] = pd.to_numrtic(df["col_name"], errors="coerce")

Add the parameter decimal=',' challenge pd.read_csvthat's better than doing. replace I don't know. Of course, it's a strange situation where you and the separator of the columns and the division of the grades have the same symbol. ,but, in principle, this may be, in this case, the contents of the columns come in quotes and Pandas I can figure out where.Either if you want to change manually, you can do it. .str.replace(',','.')And your option. .str.replace(',','.','') Changed everything into an empty line, which eventually became meaning. NA after change float♪

Any model of machine training/non-tural networks always expects data in a given format or, in technical language, a tensor of the same size, not taking into account the first or last axle (number of samples). For example, if you have a data set of 2D matrix or table, the model will always expect the same number of columns. The number of lines may vary.Data outside the sample is the data you'll get at the entrance. future years♪ In order to launch and test such a model, the following are done:A fairly large sample of data is collectedsplit into two parts - training and test sampleIn the case of a time series, the last 20 to 30 per cent of the data are usually taken as a test sample. If this is not a time series, then the data are intermittently mixed and divided into two parts - 70 to 80 per cent for training, the rest for testing.We further learn the model on the training set, and the accuracy of the model is checked on the test set.When we're satisfied with the results of the tests, the model rolls out into the market and starts working on completely new data. Of course, all the pre-processing (pre-processing) steps you've done for the study and test sample should be repeated for new input data.after this, its model is usually supplemented by new data.That's it. ♪ ♪

Is it true that I find a matrix of signs and Target variable from the datesetNo, you have a logical mistake - there's a target variable in the study sample. As a consequence, the model can be "intelligent" to pick the weight of the ov coefficient - 1.0 for a column with a target variable and zero for all other pillars. Ultimately, we'll get 100% precision for the learning sample.Separate the training sample to the input matrix and target variable. In addition, the entire sample should be divided into two parts, the training and test kits. The training model should not see the data from the data set. Otherwise, you will not be able to adequately assess the accuracy of the model for new data, i.e. for those that the model has not seen during the training:from sklearn.model_selection import train_test_split X_train, X_test, y_train, y_test = train_test_split( data.drop(columns="target"), data["target"], test_size=0.2, random_state=42)

Actually, I don't think you' matplotlib... But if my assumption is correct, try to add a parameter.tick_label: str, massive or optional o Cell for pole.challenge plt.bar(.) ♪ )♪ There's a great book: Abdrahmanov M.I. Python. Free! Issued devpractice.ru. It's very detailed on page 95.

Use the direct indication of the columns on which the association will be directed.df_2 = df_p_CG2.merge(df_CG2, left_on='user_id', right_on='user_id', how='inner') If it doesn't work, we need to look at the other frictions.

Maybe you get the type of data. float64'Cause in some lines, you're dating. NaN values and whole types in Numpy do not understand NaN and therefore Pandas automatically converts such columns into data type float64♪In modern versions, Pandas introduced a new type of data "Int64" (with a large letter) which understands the values NaN♪Example:In [13]: df = pd.DataFrame([[1, 2, 3], [np.nan, 3, np.nan]], columns=list("abc")) In [14]: df Out[14]: a b c 0 1.0 2 3.0 1 NaN 3 NaN In [17]: cols = df.columns[df.dtypes.eq("float64")] In [18]: df[cols] = df[cols].astype(int) ValueError Traceback (most recent call last) ... ValueError: Cannot convert non-finite values (NA or inf) to integer ... Decision:In [19]: df[cols] = df[cols].astype("Int64") In [20]: df Out[20]: a b c 0 1 2 3 1 <NA> 3 <NA> In [21]: df.dtypes Out[21]: a Int64 b int64 c Int64 dtype: object

The vectored Pandas / Numpy decisions, in your case, will fall sharply at the processing speed of specialized libraries that are operating. https://loginom.ru/blog/apriori And he's like that.References to modules: https://pypi.org/project/efficient-apriori/ https://borgelt.net/doc/fpgrowth/fpgrowth.html

Use the method. https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.shift.html :df[["a1","b1"]] = df[["a","b"]].shift() df[["a2","b2"]] = df[["a","b"]].shift(2) result:In [225]: df Out[225]: a b a1 b1 a2 b2 0 a0 b0 NaN NaN NaN NaN 1 a1 b1 a0 b0 NaN NaN 2 a2 b2 a1 b1 a0 b0 3 a3 b3 a2 b2 a1 b1 4 a4 b4 a3 b3 a2 b2 5 a5 b5 a4 b4 a3 b3 6 a6 b6 a5 b5 a4 b4 7 a7 b7 a6 b6 a5 b5 8 a8 b8 a7 b7 a6 b6 9 a9 b9 a8 b8 a7 b7

Comparison operators for the DataFrame's have been redesigned to the PandasExample:In [207]: d1 = pd.DataFrame([[1,2,3], [4,5,6], [7,8,9]], columns=list("abc")) In [208]: d2 = pd.DataFrame([[2,2,1], [7,8,4], [1,2,9]], columns=list("abc")) In [209]: d1 Out[209]: a b c 0 1 2 3 1 4 5 6 2 7 8 9 In [210]: d2 Out[210]: a b c 0 2 2 1 1 7 8 4 2 1 2 9 In [211]: d1 < d2 Out[211]: a b c 0 True False False 1 True True False 2 False False False In [212]: (d1 < d2).astype("int8") Out[212]: a b c 0 1 0 0 1 1 1 0 2 0 0 0

Option 1:In [81]: (pd.concat([df1, df2, df3]) .set_index("Formula") .stack() .reset_index(name="level") .drop(columns="level_1")) Out[81]: Formula level 0 \tFormula A: UNIQUES(A) 1.0 1 \tFormula A: UNIQUES(A) 2.0 2 \tFormula A: UNIQUES(A) 3.0 3 \tFormula B: UNIQUES(A) 4.0 4 \tFormula B: UNIQUES(A) 5.0 5 \tFormula B: UNIQUES(A) 6.0 6 \tFormula C: UNIQUES(A) 7.0 7 \tFormula C: UNIQUES(A) 8.0 8 \tFormula C: UNIQUES(A) 9.0 Option 2:In [84]: pd.concat(map(lambda x: x.rename(columns=lambda c: re.sub("(level_number)\d+", r"\1", c)), [df1, df2, df3])) Out[84]: Formula level_number 0 \tFormula A: UNIQUES(A) 1 1 \tFormula A: UNIQUES(A) 2 2 \tFormula A: UNIQUES(A) 3 0 \tFormula B: UNIQUES(A) 4 1 \tFormula B: UNIQUES(A) 5 2 \tFormula B: UNIQUES(A) 6 0 \tFormula C: UNIQUES(A) 7 1 \tFormula C: UNIQUES(A) 8 2 \tFormula C: UNIQUES(A) 9

decision https://ru.stackoverflow.com/questions/1342461/%D1%83%D0%B4%D0%B0%D0%BB%D0%B8%D1%82%D1%8C-%D0%B4%D1%83%D0%B1%D0%BB%D0%B8%D0%BA%D0%B0%D1%82%D1%8B-%D0%BF%D1%80%D0%B8-%D1%83%D1%81%D0%BB%D0%BE%D0%B2%D0%B8%D0%B8#comment2377468_1342461 :res = (df .assign(x=df.groupby("year")["city"].transform("nunique")) .query("""city == city or x == 0""") .drop(columns="x")) result:In [37]: res Out[37]: city year 0 Лодзь 990 1 Краков 990 3 Перм 1826 5 NaN 1827

The risk level is now determined by simple mathematical formulas.Here, take those things that go into the formulas as a fitch. I understand that every type of risk is considered to be separate, so fill out the label where these "Fitches" and the meaning of the target variable is the risk. And then take, for example, the usual linear regression and apply it to the fiction and target variable.But, most likely, if it's already considered formulae, perhaps the skill training for you won't help, will eventually end up with the same formula. But it's important for you to start getting on the subject. Read the linear regression for starters. This is, in fact, a set of coefficients for the target variable formula for different parameters (the ficcs).