Explore the curious case of Snapchat AI’s sudden story appearance. Delve into the possibilities of hacking and the true story behind the phenomenon. Curious about why your Snapchat AI suddenly has a story? Uncover the truth behind the phenomenon and put to rest concerns about whether Snapchat AI has been hacked. Explore the evolution of AI-generated stories, debunking hacking myths, and gain insights into how technology is reshaping social media experiences. Decoding the Mystery of Snapchat AI’s Unusual Story The Enigma Unveiled: Why Does My Snapchat AI Have a Story? Snapchat AI’s Evolutionary Journey Personalization through Data Analysis Exploring the Hacker Hypothesis: Did Snapchat AI Get Hacked? The Hacking Panic Unveiling the Truth Behind the Scenes: The Reality of AI-Generated Stories Algorithmic Advancements User Empowerment and Control FAQs Why did My AI post a Story? Did Snapchat AI get hacked? What should I do if I’m concerned about My AI? What is My AI...
Get link
Facebook
X
Pinterest
Email
Other Apps
Massaging Data Using Pandas
Get link
Facebook
X
Pinterest
Email
Other Apps
Last Updated on June 21, 2023
When we talk about managing data, it is pretty inevitable to see data supplied in tables. With column header, and customarily with names for rows, it makes understanding data less complicated. In actuality, it often happens that we see data of assorted varieties staying collectively. For occasion, we now have quantity as numbers and title as strings in a desk of parts for a recipe. In Python, we now have the pandas library to help us take care of tabular data.
After ending this tutorial, you will be taught:
What the pandas library provides
What is a DataPhysique and a Series in pandas
How to manage DataPhysique and Series previous the trivial array operations
Kick-start your mission with my new e ebook Python for Machine Learning, along with step-by-step tutorials and the Python provide code recordsdata for all examples.
Let’s get started!
Massaging Data Using Pandas Photo by Mark de Jong. Some rights reserved.
Overview
This tutorial is break up into 5 parts:
DataPhysique and Series
Essential capabilities in DataPhysique
Manipulating DataFrames and Series
Aggregation in DataFrames
Handling time sequence data in pandas
DataPhysique and Series
To begin, let’s start with an occasion dataset. We will import pandas and skim the U.S. air pollutant emission data proper right into a DataPhysique:
This is a desk of pollutant emissions for yearly, with the data on what kind of pollutant and the amount of emission per 12 months.
Here we demonstrated one useful attribute from pandas: You can study a CSV file using read_csv() or study an Excel file using read_excel(), as above. The filename may very well be a local file in your machine or an URL from the place the file might be downloaded. We found about this URL from the U.S. Environmental Protection Agency’s website. We know which worksheet incorporates the information and from which row the information begins, subsequently the extra arguments to the read_excel() carry out.
The pandas object created above is a DataPhysique, supplied as a desk. Similar to NumPy, data in Pandas are organized in arrays. But Pandas assign an data form to columns barely than a complete array. This permits data of assorted varieties to be included within the an identical data development. We can confirm the information form by each calling the data() carry out from the DataPhysique:
1
2
...
df.data()# print data to show display
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
<class ‘pandas.core.physique.DataPhysique’>
RangeIndex: 5319 entries, 0 to 5318
Data columns (full 32 columns):
# Column Non-Null Count Dtype
— —— ————– —–
0 State FIPS 5319 non-null int64
1 State 5319 non-null object
2 Tier 1 Code 5319 non-null int64
3 Tier 1 Description 5319 non-null object
4 Pollutant 5319 non-null object
5 emissions90 3926 non-null float64
6 emissions96 4163 non-null float64
7 emissions97 4163 non-null float64
…
29 emissions19 5052 non-null float64
30 emissions20 5052 non-null float64
31 emissions21 5052 non-null float64
dtypes: float64(27), int64(2), object(3)
memory utilization: 1.3+ MB
or we’ll moreover get the sort as a pandas Series:
1
2
3
...
coltypes=df.dtypes
print(coltypes)
1
2
3
4
5
6
7
8
9
10
11
12
13
State FIPS int64
State object
Tier 1 Code int64
Tier 1 Description object
Pollutant object
emissions90 float64
emissions96 float64
emissions97 float64
…
emissions19 float64
emissions20 float64
emissions21 float64
dtype: object
In pandas, a DataPhysique is a desk, whereas a Series is a column of the desk. This distinction is important on account of data behind a DataPhysique is a 2D array whereas a Series is a 1D array.
Similar to the flamboyant indexing in NumPy, we’ll extract columns from one DataPhysique to create one different:
State Pollutant emissions19 emissions20 emissions21
0 AL CO 8.243679 8.243679 8.243679
1 AL NH3 0.417551 0.417551 0.417551
2 AL NOX 19.592480 13.752790 11.162100
3 AL PM10-PRI 2.868642 2.868642 2.868642
4 AL PM25-PRI 2.659792 2.659792 2.659792
… … … … … …
5314 WY NOX 0.374873 0.374873 0.374873
5315 WY PM10-PRI 2.857886 2.857886 2.857886
5316 WY PM25-PRI 2.421937 2.421937 2.421937
5317 WY SO2 0.208817 0.208817 0.208817
5318 WY VOC 6.645249 6.645249 6.645249
[5319 rows x 5 columns]
Or, if we cross in a column title as a string barely than a list of column names, we extract a column from a DataPhysique as a Series:
1
2
3
...
data2023=df[“emissions21”]
print(data2023)
1
2
3
4
5
6
7
8
9
10
11
12
0 8.243679
1 0.417551
2 11.162100
3 2.868642
4 2.659792
…
5314 0.374873
5315 2.857886
5316 2.421937
5317 0.208817
5318 6.645249
Name: emissions21, Length: 5319, dtype: float64
Essential Functions in DataPhysique
Pandas is feature-rich. Many vital operations on a desk or a column are supplied as capabilities outlined on the DataPhysique or Series. For occasion, we’ll see a list of air pollution lined throughout the desk above by using:
And we’ll uncover the suggest (suggest()), regular deviation (std()), minimal (min()), and most (max()) of a sequence equally:
1
2
...
print(df[“emissions21”].suggest())
But in precise truth, we’re further attainable to utilize the describe() carry out to find a model new DataPhysique. Since the DataPhysique on this occasion has too many columns, it is larger to transpose the following DataPhysique from describe():
1
2
...
print(df.describe().T)
1
2
3
4
5
6
7
8
9
10
rely suggest std min25%50%75%max
State FIPS5319.029.03948115.6673521.0000016.00000029.00000042.00000056.000000
Indeed, the DataPhysique produced by describe() might assist us get a method of the information. From there, we’ll inform how rather a lot missing data there’s (by wanting on the rely), how the information are distributed, whether or not or not there are outliers, and so forth.
Want to Get Started With Python for Machine Learning?
Take my free 7-day e-mail crash course now (with sample code).
Click to sign-up and likewise get a free PDF Ebook mannequin of the course.
Manipulating DataPhysique and Series
Similar to the Boolean indexing in NumPy, we’ll extract a subset of rows from a DataPhysique. For occasion, that’s how we’ll select the information for carbon monoxide emissions solely:
1
2
3
...
df_CO=df[df[“Pollutant”]==“CO”]
print(df_CO)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
State FIPS State Tier 1 Code Tier 1 Description … emissions18 emissions19 emissions20 emissions21
As chances are high you will anticipate, the == operator compares each issue from a sequence df["Pollutant"] , resulting in a sequence of Boolean. If the lengths match, the DataPhysique understands it is to select the rows based totally on the Boolean price. In actuality, we’ll combine Booleans using bitwise operators. For occasion, that’s how we select the rows of carbon monoxide emissions as a consequence of freeway autos:
If you want to select rows like a Python document, chances are high you will accomplish that by the use of the iloc interface. This is how we’ll select rows 5 to 10 (zero-indexed) or columns 1 to 6 and rows 5 to 10:
1
2
3
...
df_r5=df.iloc[5:11]
df_c1_r5=df.iloc[5:11,1:7]
If you’re accustomed to Excel, you most likely know one among its thrilling choices generally known as a “pivot table.” Pandas lets you do the an identical. Let’s take note of the air air pollution of carbon monoxide from all states in 2023 from this dataset:
The pivot_table() carry out above would not require the values to be distinctive to the index and columns. In completely different phrases, must there be two “wildfire” rows in a state throughout the genuine DataPhysique, this carry out will combination the two (the default is to take the suggest). To reverse the pivot operation, we now have the soften() carry out:
There is much extra we’ll do with a DataPhysique. For occasion, we’ll sort the rows (using the sort_values() carry out), rename columns (using the rename() carry out), take away redundant rows (drop_duplicates() carry out), and so forth.
In a machine finding out mission, we ceaselessly should do some clean-up sooner than we’ll use the information. It is useful to utilize pandas for this goal. The df_pivot DataPhysique we merely created has some values marked as NaN for no data accessible. We can substitute all these with zero with any of the subsequent:
1
2
3
df_pivot.fillna(0)
df_pivot.the place(df_pivot.notna(),0)
df_pivot.masks(df_pivot.isna(),0)
Aggregation in DataFrames
In actuality, pandas can current desk manipulation that in some other case can solely be merely carried out using database SQL statements. Reusing the above occasion dataset, each pollutant throughout the desk is broken down into completely completely different sources. If we have to know the aggregated pollutant emissions, we’ll merely sum up the entire sources. Similar to SQL, it’s a “group by” operation. We can accomplish that with the subsequent:
The outcomes of the groupby() carry out will use the grouping column as a result of the row index. It works by inserting rows which have the an identical price for that grouping column right into a bunch. Then as a bunch, some combination carry out is utilized to chop again the varied rows into one. In the above occasion, we’re taking the sum all through each column. Pandas comes with many various combination capabilities, just like taking the suggest or just counting the number of rows. Since we’re doing sum(), the non-numeric columns are dropped from the output as they do not apply to the operation.
This permits us to do some fascinating duties. Let’s say, using the information throughout the DataPhysique above, we create a desk of the general emission of carbon monoxide (CO) and sulfur dioxide (SO2) in 2023 in each state. The reasoning on how to do that may very well be:
Group by “State” and “Pollutant,” then sum up each group. This is how we get the general emission of each pollutant in each state.
Select solely the column for 2023
Run pivot desk to make states the rows and the air pollution the columns with the general emission as a result of the values
In the above code, each step after the groupby() carry out is to create a model new DataPhysique. Since we’re using capabilities outlined beneath DataPhysique, we now have the above sensible chained invocation syntax.
The sum() carry out will create a DataPhysique from the GroupBy object that has the grouped columns “State” and “Pollutant” as an index. Therefore, after we diced the DataPhysique to only one column, we used reset_index() to make the index as columns (i.e., there’ll in all probability be three columns, State, Pollutant, and emissions21). Since there’ll in all probability be further air pollution than we might like, we use filter() to select solely the columns for CO and SO2 from the following DataPhysique. This is rather like using fancy indexing to select columns.
Indeed, we’ll do the an identical otherwise:
Select solely the rows for CO and compute the general emission; select solely the information for 2023
Do the an identical for SO2
Combine the following DataPhysique throughout the earlier two steps
In pandas, there is a be a part of() carry out in DataPhysique that helps us combine the columns with one different DataPhysique by matching the index. In code, the above steps are as follows:
The be a part of() carry out is restricted to index matching. If you’re accustomed to SQL, the JOIN clause’s equal in pandas is the merge() carry out. If the two DataFrames we created for CO and SO2 have the states as a separate column, we’ll do the an identical as follows:
The merge() carry out in pandas can do every kind of SQL joins. We can match completely completely different columns from a particular DataPhysique, and we’ll do left be a part of, correct be a part of, inside be a part of, and outer be a part of. This will in all probability be very useful when wrangling the information in your mission.
The groupby() carry out in a DataPhysique is very efficient as a result of it permits us to manage the DataPhysique flexibly and opens the door to many refined transformations. There may be a case that no built-in carry out might assist after groupby(), nevertheless we’ll always current our private. For occasion, that’s how we’ll create a carry out to operate on a sub-DataPhysique (on all columns in addition to the group-by column) and apply it to look out the years of minimal and most emissions:
The apply() carry out is the ultimate resort to provide us the utmost flexibility. Besides GroupBy objects, there are moreover apply() interfaces in DataFrames and Series.
The following is the entire code to exhibit all operations we launched above:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
import pandas aspd
# Pollutants data from Environmental Protection Agency
You will uncover one different extremely efficient attribute from pandas for those who’re dealing with time sequence data. To begin, let’s take note of some every day air air pollution data. We can select and procure some from the EPA’s website:
For illustration capabilities, we downloaded the PM2.5 data of Texas in 2023. We can import the downloaded CSV file, ad_viz_plotval_data.csv, as follows:
The read_csv() carry out from pandas permits us to specify some columns as a result of the date and parse them into datetime objects barely than a string. This is essential for added processing time sequence data. As everyone knows, the first column (zero-indexed) is the date column; we provide the argument parse_dates=[0] above.
For manipulating time sequence data, it is vitally vital use time as an index in your DataPhysique. We may make one in all many columns an index by the set_index() carry out:
1
2
3
...
df_pm25=df.set_index(“Date”)
print(df_pm25)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Source Site ID POC Daily Mean PM2.5Concentration...COUNTY_CODE COUNTY SITE_LATITUDE SITE_LONGITUDE
We know its form is datetime64, which is a timestamp object in pandas.
From the index above, we’ll see each date is not going to be distinctive. This is on account of the PM2.5 focus is observed in a number of web sites, and each will contribute a row to the DataPhysique. We can filter the DataPhysique to only one web site to make the index distinctive. Alternatively, we’ll use pivot_table() to transform the DataPhysique, the place the pivot operation ensures the following DataPhysique can have distinctive index:
1
2
3
4
5
6
df_2023=(df[[“Date”,“Daily Mean PM2.5 Concentration”,“Site Name”]]
.pivot_table(index=“Date”,
columns=“Site Name”,
values=“Daily Mean PM2.5 Concentration”)
)
print(df_2023)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Site Name Amarillo A&M Ascarate Park SE Austin North Hills Drive … Von Ormy Highway 16 Waco Mazanec World Trade Bridge
Date …
2023-01-01 1.7 11.9 3.0 … 4.6 2.7 4.4
2023-01-02 2.2 7.8 6.1 … 3.7 2.2 6.1
2023-01-03 2.5 4.2 4.3 … 6.3 4.2 8.5
2023-01-04 3.7 8.1 3.7 … 6.4 4.2 5.7
2023-01-05 4.5 10.0 5.2 … 7.7 6.1 7.1
… … … … … … … …
2023-12-27 1.9 5.8 11.0 … 13.8 10.5 15.7
2023-12-28 1.8 6.6 14.1 … 17.7 9.7 17.6
2023-12-29 NaN 8.1 21.8 … 28.6 12.5 14.1
2023-12-30 4.0 9.5 13.1 … 20.4 13.4 18.5
2023-12-31 3.6 3.7 16.3 … 18.3 11.8 21.5
[365 rows x 53 columns]
We can confirm the individuality with:
1
df_2023.index.is_unique
Now, every column on this DataPhysique is a time sequence. While pandas would not current any forecasting carry out on the time sequence, it comes with devices that may help you clear and rework the information. Setting a DateTimeIndex to a DataPhysique will in all probability be useful for time sequence analysis initiatives on account of we’ll merely extract data for a time interval, e.g., the train-test break up of the time sequence. Below is how we’ll extract a 3-month subset from the above DataPhysique:
1
df_3month=df_2023[“2023-04-01”:“2023-07-01”]
One typically used carry out in a time sequence is to resample the information. Considering the every day data on this DataPhysique, we’ll rework it into weekly observations instead. We can specify the following data to be listed on every Sunday. But we nonetheless have to tell what we want the resampled data to be like. If it is product sales data, we most likely have to sum over your full week to get the weekly revenue. In this case, we’ll take the everyday over per week to straightforward out the fluctuations. An completely different is to take the first commentary over each interval, like underneath:
1
2
3
...
df_resample=df_2023.resample(“W-SUN”).first()
print(df_resample)
1
2
3
4
5
6
7
8
9
10
11
Site Name AmarilloA&MAscarate Park SE Austin North Hills Drive...Von Ormy Highway16Waco Mazanec World Trade Bridge
Date...
2023–01–031.711.93.0...4.62.74.4
2023–01–103.78.13.7...6.44.25.7
2023–01–175.85.37.0...5.46.94.8
...
2023–12–193.613.06.3...6.95.95.5
2023–12–265.310.45.7...5.55.43.9
2023–01–021.95.811.0...13.810.515.7
[53rowsx53columns]
The string “W-SUN” is to seek out out the suggest weekly on Sundays. It generally known as the “offset alias.” You can uncover the document of all offset alias from underneath:
Resampling is particularly useful in financial market data. Imagine if we now have the value data from the market, the place the raw data would not can be found in frequent intervals. We can nonetheless use resampling to rework the information into frequent intervals. Because it is so typically used, pandas even provides you the open-high-low-close (commonly known as OHLC, i.e., first, most, minimal, and remaining observations over a interval) from the resampling. We exhibit underneath tips about the best way to get the OHLC over per week on one in all many commentary web sites:
1
2
df_ohlc=df_2023[“San Antonio Interstate 35”].resample(“W-SUN”).ohlc()
print(df_ohlc)
1
2
3
4
5
6
7
8
9
10
11
open extreme low shut
Date
2023-01-03 4.2 12.6 4.2 12.6
2023-01-10 9.7 9.7 3.0 5.7
2023-01-17 5.4 13.8 3.0 13.8
2023-01-24 9.5 11.5 5.7 9.0
…
2023-12-12 5.7 20.0 5.7 20.0
2023-12-19 9.7 9.7 3.9 3.9
2023-12-26 6.1 14.7 6.0 14.7
2023-01-02 10.9 23.7 10.9 16.3
In express, if we resample a time sequence from a coarser frequency proper right into a finer frequency, it is generally known as upsampling. Pandas usually inserts NaN values all through upsampling because the distinctive time sequence would not have data all through the in-between time conditions. One method to steer clear of these NaN values all through upsampling is to ask pandas to forward-fill (carry over values from an earlier time) or back-fill (using values from a later time) the information. For occasion, the subsequent is to forward-fill the every day PM2.5 observations from one web site into hourly:
1
2
3
...
series_ffill=df_2023[“San Antonio Interstate 35”].resample(“H”).ffill()
print(series_ffill)
1
2
3
4
5
6
7
8
9
10
11
12
13
Date
2023-01-01 00:00:00 4.2
2023-01-01 01:00:00 4.2
2023-01-01 02:00:00 4.2
2023-01-01 03:00:00 4.2
2023-01-01 04:00:00 4.2
…
2023-12-30 20:00:00 18.2
2023-12-30 21:00:00 18.2
2023-12-30 22:00:00 18.2
2023-12-30 23:00:00 18.2
2023-12-31 00:00:00 16.3
Freq: H, Name: San Antonio Interstate 35, Length: 8737, dtype: float64
Besides resampling, we’ll moreover rework the information using a sliding window. For occasion, underneath is how we’ll make a 10-day transferring frequent from the time sequence. It is not going to be a resampling on account of the following data stays to be every day. But for each data stage, it is the suggest of the earlier 10 days. Similarly, we’ll uncover the 10-day regular deviation or 10-day most by making use of a particular carry out to the rolling object.
1
2
3
...
df_mean=df_2023[“San Antonio Interstate 35”].rolling(10).suggest()
print(df_mean)
1
2
3
4
5
6
7
8
9
10
11
12
13
Date
2023-01-01 NaN
2023-01-02 NaN
2023-01-03 NaN
2023-01-04 NaN
2023-01-05 NaN
…
2023-12-27 8.30
2023-12-28 9.59
2023-12-29 11.57
2023-12-30 12.78
2023-12-31 13.24
Name: San Antonio Interstate 35, Length: 365, dtype: float64
To current how the distinctive and rolling frequent time sequence differs, underneath displays you the plot. We added the argument min_periods=5 to the rolling() carry out on account of the distinctive data has missing data on some days. This produces gaps throughout the every day data, nevertheless we ask that the suggest nonetheless be computed as long as there are 5 data components over the window of the earlier 10 days.
1
2
3
4
5
6
7
8
9
...
import matplotlib.pyplot asplt
fig=plt.decide(figsize=(12,6))
plt.plot(df_2023[“San Antonio Interstate 35”],label=“every day”)
plt.plot(df_2023[“San Antonio Interstate 35”].rolling(10,min_periods=5).suggest(),label=“10-day MA”)
plt.legend()
plt.ylabel(“PM 2.5”)
plt.current()
The following is the entire code to exhibit the time sequence operations we launched above:
In this tutorial, you observed a fast overview of the capabilities supplied by pandas.
Specifically, you found:
How to work with pandas DataFrames and Series
How to manage DataFrames in a way similar to desk operations in a relational database
How to make the most of pandas to help manipulate time sequence data
Get a Handle on Python for Machine Learning!
Be More Confident to Code in Python
…from finding out the smart Python suggestions
Discover how in my new Ebook: Python for Machine Learning
It provides self-study tutorials with a complete bunch of working code to equip you with experience along with: debugging, profiling, duck typing, decorators, deployment, and quite extra…
Showing You the Python Toolbox at a High Level for Your Projects
See What’s Inside
More On This Topic
Multi-Step LSTM Time Series Forecasting Models for…
How to Perform Data Cleaning for Machine Learning…
Deep Learning Models for Univariate Time Series Forecasting
Python Machine Learning Mini-Course
Convolutional Neural Networks for Multi-Step Time…
Raising your deductible and Older motorcar – Drop comprehensive / collision coverage. Deductible is that the quantity you pay from your pocket before creating associate claim. The disadvantage of raising your claim is once you build a claim, you’ll pay more. However, if you’re a secure driver, you will overtime save more cash by raising your insurance deductible. cross-check your previous insurance claim history and make a discreet call for yourself. If your automotive isn’t value much, why get comprehensive and collision insurance coverage. you’ll be {able to} visit a myriad of on-line sites to search out true worth of your car. to boot your factor may well be able to pull up truth worth of your vehicle. Taking advantage of low mileage and Moving – take into account insurance costs. Some motorcar insurance firms can provide discounts if you drive under a particular variety of miles or drive less than a certain distance to work. If you’re cons...
Last Updated on September 22, 2023 Sponsored Post 📣 The AI event of the 12 months is quickly approaching… We’re talking about TransformX , a FREE digital conference the place you’ll hear from 120+ know-how leaders from companies like Google, Meta, OpenAI, DeepMind, Amazon, and additional. Explore how AI will vitality ecommerce, AI capabilities for healthcare, NFT marketplaces and additional. 🎙 Speakers embrace: Jeff Wilke – Retired as Amazon’s CEO Worldwide Consumer Dr. Kenneth E. Washington – Vice President of Software Engineering at Amazon Eric Schmidt – Co-Founder Schmidt Futures, Former CEO at Google James Manyika – Senior Vice President for Technology & Society at Google John List – Kenneth C. Griffin Distinguished Service Professor in Economics, University of Chicago Greg Brockman – President, Chairman and Founder at OpenAI Koray kavukcuoglu – VP of Research and Technology at DeepMind Laura Major – CTO at Motional + many further! Grab a ticket today, it’s fr...
Explore the curious case of Snapchat AI’s sudden story appearance. Delve into the possibilities of hacking and the true story behind the phenomenon. Curious about why your Snapchat AI suddenly has a story? Uncover the truth behind the phenomenon and put to rest concerns about whether Snapchat AI has been hacked. Explore the evolution of AI-generated stories, debunking hacking myths, and gain insights into how technology is reshaping social media experiences. Decoding the Mystery of Snapchat AI’s Unusual Story The Enigma Unveiled: Why Does My Snapchat AI Have a Story? Snapchat AI’s Evolutionary Journey Personalization through Data Analysis Exploring the Hacker Hypothesis: Did Snapchat AI Get Hacked? The Hacking Panic Unveiling the Truth Behind the Scenes: The Reality of AI-Generated Stories Algorithmic Advancements User Empowerment and Control FAQs Why did My AI post a Story? Did Snapchat AI get hacked? What should I do if I’m concerned about My AI? What is My AI...
Comments
Post a Comment