MIT Introduction to Deep Learning 6.S191: Lecture 1 *New 2020 Edition*

MIT Introduction to Deep Learning 6.S191: Lecture 1 *New 2020 Edition*


https://www.youtube.com/watch?v=njKP3FqW3Sk&list=PLtBw6njQRU-rwp5__7C0oIVt26ZgjG9NI&index=2

MIT Introduction to Deep Learning 6.S191: Lecture 1 *New 2020 Edition* Foundations of Deep Learning Lecturer: Alexander Amini January 2020 For all lectures, slides, and lab materials: http://introtodeeplearning.com Lecture Outline 0:00 - Introduction 4:14 - Course information 8:10 - Why deep learning? 11:01 - The perceptron 13:07 - Activation functions 15:32 - Perceptron example 18:54 - From perceptrons to neural networks 25:23 - Applying neural networks 28:16 - Loss functions 31:14 - Training and gradient descent 35:13 - Backpropagation 39:25 - Setting the learning rate 43:43 - Batched gradient descent 46:46 - Regularization: dropout and early stopping 51:58 - Summary Subscribe to @stay up to date with new deep learning lectures at MIT, or follow us on @MITDeepLearning on Twitter and Instagram to stay fully-connected!!

Pandas cheat sheet - Шпаргалка панд

Pandas cheat sheet
Шпаргалка панд

https://morphocode.com/pandas-cheat-sheet/

Pandas cheat sheet

Data can be messy: it often comes from various sources, doesn’t have structure or contains errors and missing fields. Working with data requires to clean, refine and filter the dataset before making use of it.

Pandas is one of the most popular tools to perform such data transformations. It is an open source library for Python offering a simple way to aggregate, filter and analyze data. The library is often used together with Jupyter notebooks to empower data exploration in various research and data visualization projects.

Pandas introduces the concept of a DataFrame – a table-like data structure similar to a spreadsheet. You can import data in a data frame, join frames together, filter rows and columns and export the results in various file formats. Here is a pandas cheat sheet of the most common data operations:

Getting Started

Import Pandas & Numpy

import numpy as np

import pandas as pd

Get the first 5 rows in a dataframe:

df.head(5)

Get the last 5 rows in a dataframe:

df.tail(5)

Import Data

Create DataFrame from dictionary:

df = pd.DataFrame.from_dict({

    'company': 'Pandology', 

    'metrics': [[{'score': 10}, {'score': 20}, {'score': 35}]]

})

Import data from a CSV file:

df = pd.read_csv('data/my-data.csv')

Import data from an Excel Spreadsheet:

df_excel = pd.read_excel('./data/spreadsheet.xls',

    sheetname='sheet1',

    skiprows=[1] # header data

)

Import data from an Excel Spreadsheet without the header:

df_names = pd.read_excel('./data/names_all.xls', header=None)

Export Data

Export as an Excel Spreadsheet:

df[['name_company', 'name_candidate']].to_excel('./output/companies.xls')

Export to a CSV file:

df.to_csv('data-output/my-data.csv')

Convert Data Types

Convert column data to string:

df['name'] = df['name'].astype('str')

Convert column data to integer (nan values are set to -1):

df['col'] = df['col'].fillna(-1).astype(int)

Convert column data to numeric type:

df['col'] = pd.to_numeric(df['col'])

Get / Set Values

Get the value of a column on a row with index idx:

df.get_value(idx, 'col_name')

Set column value on a given row:

idx = df[df['address'] == '4th Avenue'].index

df.set_value(idx, 'id', '502')

Count

Number of rows in a DataFrame:

len(df)

Count rows where column is equal to a value:

len(df[df['score'] == 1.0])

Count unique values in a column:

df['name'].nunique()

Count rows based on a value:

df['is_removed'].value_counts()



# Count null values as well:

df['is_removed'].value_counts(dropna=False)

Filter Data

Filter rows based on a value:

df[df['id'] == '48']

Filter rows based on multiple values:

df[(df['category'] == 'national') & (df['is_removed'] == '1')]

Filter rows that contain a string:

df[df['category'].str.contains('national')]

Filter rows containing some of the strings:

df['address'].str.contains('|'.join(['4th Avenue', 'Broadway']))

Filter rows where value is in a list:

df[df['id'].isin(['109', '673'])]

Filter rows where value is _not_ in a list:

df = df[~df['id'].isin(['1', '2', '3'])]

Filter all rows that have valid values (not null):

df = df[pd.notnull(df['latitude'])]

Sort Data

Sort rows by value:

df.sort_values('nom', inplace=True)

Sort Columns By Name:

df = df.reindex_axis(sorted(df.columns), axis=1)

Rename columns

Rename particular columns:

df.rename(columns={'id': 'id_new', 'object': 'object_new'}, inplace=True)

Rename all columns:

df.columns = ['id', 'object', 'address', 'type', 'category']

Make all columns lowercase:

df.columns = map(str.lower, df.columns)

Drop data

Drop column named col

df = df.drop('col', axis=1)

Drop all rows with null index:

df[pd.notnull(df.index)]

Drop rows that have missing values in some columns:

df.dropna(subset=['plsnkv', 'plsnnum'])

Drop duplicate rows:

df = df.drop_duplicates(subset='id', keep='first')

Create columns

Create a new column based on row data:

def cad_id(row):    

    return row['region'] + '.' + row['lot'] + '.' + row['building']

      

df['cad_id'] = df.apply(cad_id, axis=1)

Create a new column based on another column:

df['is_removed'] = df['object'].map(lambda x: 1 if 'removed' in x else 0)

Create multiple new columns based on row data:

def geocode(row):

    address = api_geocode(street=row['street'], house_number=row['building'], zip_code=row['zip'])   

    return pd.Series([address.get('lat'), address.get('lon'), address.get('borough')])

     

df[['lat', 'lon', 'borough']] = df.apply(geocode, axis=1)

Match id to label:

def zone_label(zone_id):

    return {

        'C': 'City Center',

        'S': 'Suburbs',       

        'R': 'Residential District'

    }.get(zone_id, zone_id)   

df['zone_label'] = df['zone_id'].map(zone_label)

Data Joins

Join data frames by columns:

df_lots.merge(df_buildings, on=['lot_id', 'mun_id'], suffixes=('_lot', '_building'))

Concatenate two data frames (one after the other):

df_all = pd.concat([df_first, df_second])

Utilities

Increase the number of table rows & columns shown:

pd.options.display.max_rows = 999 

pd.options.display.max_columns = 999