FRIENDLY Archives - Artificial Intelligence

Complete Guide To Arrow: A Python Library for User-friendly Handling Of Dates, Time and Timestamps

aiuniverse — Fri, 19 Mar 2021 06:45:21 +0000

Source – https://analyticsindiamag.com/

Arrow is a flexible Python library designed to create, format, manipulate, and convert dates, time, and timestamps in a sensible and human-friendly manner. It provides an intelligent module API that allows dealing with dates and times with a few code lines and imports. The library is named after the “time’s arrow” concept suggesting the one-directional progress of time. It is inspired by Requests (an HTTP library) and Moment.js (a JavaScript date library).

Why Arrow?

Standard Python library and low-level conventional modules are available for the various date and time functionalities but have the following limitations from a user’s point of view:

Too many modules available such as time, datetime, dateutil, calendar and pytz
A large number of data types to choose from such as datetime, date, time, timedelta, tzinfo and so on
Users may lack experience in handling different time zones
Inefficient ways of timestamps and timezone conversions
Gaps in functionalities such as humanization (e.g. converting a date into human-readable format) and ISO 8601 parsing

Highlighting Features of Arrow

Supports Python 3.6+ versions
A fully-implemented, efficient alternative to datetime standard library
Aware of timezones (uses UTC (Coordinated Universal Time) by default) and enables easy timezone conversion.
Automatic formatting and parsing of strings
Widely supports ISO 8601 standard date and time format.
Supports pytz, ZoneInfo and dateutil objects
Can generate floors, ceilings, spans and ranges for time frames ranging from microseconds to years
Can be extended to users’ own Arrow-derived datatypes
Supports PEP 484-style type hints

Installing Arrow

Install Arrow using pip command as follows:

pip install -U arrow

Practical Implementation

Here’s a demonstration of performing various operations using Arrow. The experiments have been done in Google colab with Python version3.7.10. Explanation of each operation along with its output is as follows:

First, import the Arrow library

import arrow as arw

Represent current time in the given timezone

arw.now()

With no timezone specified, an Arrow object representing current UTC time will be created. Same output results on executing

arw.utcnow()

Output:

To get current time in say US/Pacific timezone:

arw.now(‘US/Pacific’)

Output:

Convert a specified integer or float number into a floating-point UTC timestamp

arw.get(1567900664)

Output:

 #Try with a floating point input
 arw.get(17900664.5463877)

Output:

Verify the converted output here.

String parsing

arw.get('12-03-2020 22:30:25', 'MM-DD-YYYY HH:mm:ss')

Specified date and time will be represented as MM-DD-YYYY HH:mm:ss

format

Output:

Extract date from a string

arw.get('I was born on March 12 1990','MMMM DD YYYY')

‘MMMM DD YYYY’ formatted date will be searched in the string

Output:

Instantiate Arrow object by directly providing a datetime argument

arw.Arrow(2020,12,26)

Output:

Get a datetime representation an Arrow object

x = arw.now()

Suppose, x is

x.datetime

Output: datetime.datetime(2021, 3, 16, 11, 49, 17, 908895, tzinfo=tzlocal())

The components can then be separated as:

x.month

Output: 3

x.year

Output: 2021

x.day

Output: 16

x.hour

Output: 11

Datetime functions can be called to get properties of the Arrow object

x.time()

Output: datetime.time(11, 49, 17, 908895)

Replace one or more components of the Arrow object

a = arw.now()

Suppose, ‘a’ is

a.replace(year=2019, second=45, month=11)

will give the output:

One or more Arrow object attributes can be shifted forward or backward

present = arw.now()

Suppose, ‘present’ object is

Go forward in time by 2 years

present.shift(years=+2)

Output:

Go backward by 4 hours

present.shift(hours=-4)

Output:

Represent date and time in the required format

arw.now().format('HH:MM:SS MM:DD:YYYY')

Output: 12:03:00 03:16:2021

(i.e. 12 hrs 3 mins 00 sec, 16 March 2021)

Convert default UTC to specified timezone

utc = arw.utcnow()

Suppose, utc is

It can be converted to US/Pacific time zone as follows:

utc.to('US/Pacific')

Output:

Humanization examples

Humanize relative to current time:

 future = arw.now().shift(minutes=+2)  #advance by 2 mins
 future.humanize()

Output: ‘in 2 minutes’

Humanize relative to another datetime or Arrow object

 present = arw.now()  #current time
 future = present.shift(days=-3)  #go back by 3 days
 #how far is present from future
 future.humanize(present)

Output: ‘3 days ago’

 #how far is future from present
 present.humanize(future)

Output: ‘in 3 days’

To get only the distance:

future.humanize(present, only_distance=True)

Output: ‘3 days’

Indicate time difference in terms of one or more specific time granularities like hours,

minutes etc.

future.humanize(present, granularity="minute")

Output: ‘4320 minutes ago’ #3 days equals 4320 mins

Multiple granularities can be specified as:

future.humanize(present, granularity=["minute","second"])

Output: ‘in 4320 minutes and 0 seconds’

Get the span of any time unit

arw.now().span(‘minute’)

Output:

 (,
  )

arw.now().span('day') #get the span of time for a whole day

Output:

 (,
  )

Get floor and ceiling values of a specific time component

arw.now().floor('minute')

Output:

arw.now().ceil('minute')

Output:

Get the range of specified timespans

 import datetime
 begin = datetime.datetime(2020, 6, 5, 2, 30) #start time
 end = datetime.datetime(2020, 6, 5, 4, 50)  #end time
 for r in arw.Arrow.range('hour', begin, end):  #iterate in terms of hour
     print(r)

We want time from 2:30 to 4:50 at the interval of an hour so output will contain 2:30, 3:30 and 4:30 times.

Output:

 2020-06-05T02:30:00+00:00
 2020-06-05T03:30:00+00:00
 2020-06-05T04:30:00+00:00

FACTORIES can be used for utilizing Arrow’s module API to create a custom Arrow-derive type.

Suppose, we need to find difference in terms of days between today and the Christmas day. We can define a custom class as:

 class Custom(arw.Arrow):  #pass an Arrow object
      def till_christmas(self):  #define a function for computation
   #store christmas day of given year
          christmas = arw.Arrow(self.year, 12, 25)
 “””
 If given date falls comes after Christmas of that year, compute its difference w.r.t. Christmas of the next year
 “””
          if self > christmas:
              christmas = christmas.shift(years=1)
          return (christmas - self).days  #return the difference

 f = arw.ArrowFactory(Custom) #Create a factory
 x = f.now()  #current date and time

Suppose, x is

Call the till_christmas() method to compute the difference

x.till_christmas()

Output: 283

i.e. Christmas of the year 2021 is 283 days away from 16 March 2021.

The post Complete Guide To Arrow: A Python Library for User-friendly Handling Of Dates, Time and Timestamps appeared first on Artificial Intelligence.

DATA SCIENCE: UPGRADING ACCESSIBILITY THROUGH USER-FRIENDLY APPROACHES

aiuniverse — Fri, 12 Mar 2021 09:29:14 +0000

Source – https://www.analyticsinsight.net/

Organizations should consider democratizing data science to enhance its benefits to a broader audience.

Data science is the backbone of many organizations in the current scenario. Businesses are continuously competing to become data-driven to enhance agility and growth especially in today’s time of crisis. The pandemic accelerated digital transformation across the industries and data plays a huge role in achieving this transformation. According to BARC research, organizations leveraging Big data reported an average 8% increase in revenues and a 10% decrease in costs.

Data science for business is not a new strategy and thus is not unfamiliar. Artificial intelligence and data science go hand in hand to automate operations and deliver the best business insights. Data is a critical business asset and hence data science has a great impact on revenue, operation costs, risk management, customer experience, supply chain and logistics management, predictive analytics, and fraud detection.

To achieve business productivity and growth, many organizations are investing in data science by hiring data scientists into their systems. Data scientists are professionally qualified in studying data and deriving insights from them. However, due to the importance of data science, companies tend to centralize on data scientists rather, which decreases the awareness among the others.

Data science needs to be decentralized and democratizing data science in organizations should be an important business agenda.

Democratizing Data Science- As a Strategy

Extending the accessibility of data science across an organization needs a definite plan. For example, understanding and implementing data science for beginners might be difficult considering the lack of communication and awareness on the subject.

• Once they have data scientists, companies often forget the need to communicate the broad vision of data science to others. This scenario should stop. Enhancing communication and educating employees on the power of data and data science is a crucial step in democratizing data science in organizations. Employers should work towards upskilling their employees on data analytics and gaining business insights from it. This kind of collaboration within a work environment will increase the knowledge about data science and enable people other than data scientists to leverage data science. Involvement is the key.

• Unburdening your data scientists can be another way to standardize data science in your organization. Most of the organizations maintain their data science tools with the specialized data science team creating a silo by restricting other employees from accessing these tools. The data science silo needs to be broken so that these tools can be shared among others enabling them to act on issues that need basic training. Thus, the data scientist team can focus on tasks that need high-level expertise. Integrating AI and machine learning systems to automate repetitive tasks can also free up data scientists.

• Sharing tools are not enough since analyzing data requires specific skills. Organizations should make sure that they create a knowledge pool to share data science skills and train employees on the same. Once the employees have hands-on experience, it becomes easier to find solutions. When employees start understanding the language of data the business productivity escalates considering the huge capabilities of data analytics and science.

• Organizations should incorporate user-friendly tools like no-code or low-code automation, graphical user interfaces, drag and drop technology, etc. to enhance the accessibility of data analytics to the whole organization.

Data science in organizations should be accessible to all the employees rather than restricting it to a particular group. The strategy of democratizing data science in organizations will enhance the scalability, enable better innovations, fast-track productivity, and minimize costs.

An article by Jeff Feng, PM lead for Data at Airbnb reveals how the organization has embraced data science democratization by introducing Data University, which is ‘data education for anyone.’ He says, “Our vision is to empower every employee to make data-informed decisions. Our approach is unique since organizations offering data education typically focus just on their technical employees. Our approach is also intentional because we believe that every person at Airbnb should and can utilize data in his/her role to make better decisions. Thus, we designed the program to make it accessible and relevant to anyone at Airbnb.”

The post DATA SCIENCE: UPGRADING ACCESSIBILITY THROUGH USER-FRIENDLY APPROACHES appeared first on Artificial Intelligence.