# To add a new cell, type '#%%' # To add a new markdown cell, type '#%% [markdown]' #%% [markdown] # # # HW OOP # ## By: xxx # ###

Date: xxxxxxx # #%% [markdown] # # Part I # # Remember toggle between line wrap, (Option-Z on Mac, and Alt-Z on Windows?) to view # longer lines. I like to keep some lines intact, to keep the overall structure easier # to follow. # # Let us try to use OOP to manage something useful. # # This first part of the exercise, we will use just basic python and OOP to # create a Stock class, to keep track of different stocks in a portfolio. # Let's say, Apple. (NO numpy, pandas or anything fancy here in this homework.) # # I pulled data from https://old.nasdaq.com/symbol/aapl/historical (5 years, # csv format, 9/12/2019). You can also try from something like www.quandl.com/EOD # # The csv file is given to you here, along with Microsoft (MSFT) and Google (the parent # company is Alphabet now, stock symbol GOOG) files. # # Our goal: Pull in the basic data, store it into a Stock class object. # Next, derive some useful numbers and functions out of it. # For example, from the daily data with only date, end-of-day closing price, and # trading volumn as three columns, we want to be able to find the daily changes easily, # and also the change of the last 50 days, 100 days, for example. # The daily change represents the "first derivatives" in calculus or economics term. # It is also very common to look at the "second derivatives", which is the "rate of # change of the rate of change of the stock price". This "second derivative" represent # something like a overall trend/momentum. A negative 2nd derivative means even though # the stock price change (1st derivative) is positive, but the increase is slowing down. # You might want to find a time to get out. # # So this is the structure of my proposed Stock class: # Attributes: # :param symbol: (string) stock symbol # :param name: (string) company name # :param firstdate: (python list) the first date (end of list) of the price list # :param lastdate: (python list) the last date (beginning of list) of the price list # :param init_filepath: locate the file for date, price (eod) and volume lists to initialize this stock object # below is other attributes that will be kept as a python list. When the object is initialized, # it reads in the data file and calculate the rest # :self.dates: = [] # (python list) starts from the latest/newest date, for easy access of newer data when the list gets long # :self.price_eod: = [] # (python list) import from data file # :self.volumes: = [] # (python list) import from data file # :self.delta1: = [] # (python list) daily change values, or the first derivative, that is the previous close minus today's close, example price_eod[0] - price_eod[1], price_eod[1] - price_eod[2] # :self.delta2: = [] # (python list) the second derivative, calculate from teh first derivative. Basicallly, the daily change in the first derivative (over 1 day), is the second derivative. # # Methods: # import_history # import from csv file, then populate the corresponding values for the attributes # compute_delta1_list # from the price_eod (a python list), calculate a list of daily changes. Notice that if your eod list has length n, you will only get a list of length n-1 for this list. # compute_delta2_list # similarly calculated from delta1 list. And if your delta1 list has length n-1, you will only get a list of length n-2 here. # insert_newday # Every day, I will need to insert a new row of data to keep up-to-date. This task involves adding values to eod, dates, and volumn lists, and also need to calculate and add new values to delta1 and delta2. # nday_change_percent # Need to calculate the percent change of price in the last n-days. # # # Most of the codes are written for you. Your task is to understand them, and fill in the rest # to make it work. # # Let's start: #%% # Step 0, try reading the data file to make sure it works. If your data file is in the same # folder, you don't need to put in the full path. Just the file name. # # Run this cell as is. If you see some output in the interactive python window, then it is # working. If not, you might need to fix the file path accordingly for your OS/platform. # You can use functions like os.getcwd() to find out what is the current working directory # in your ipython session. If needed, you can use os.chdir( your path ) to set the current # directory to where you need. like: # filepath = "/Users/edwinlo/GDrive_GWU/github_elo/GWU_classes_p/DATS_6103_DataMining/Assignment/AAPL_daily.csv" import os appl_date = [] appl_price_eod = [] filepath = os.path.join( os.getcwd(), "AAPL_daily.csv") fh = open(filepath) # fh stands for file handle # data pulled from https://old.nasdaq.com/symbol/aapl/historical (5 years, csv format, 9/12/2019) can also try www.quandl.com/EOD for aline in fh.readlines(): # readlines creates a list of elements; each element is a line in the txt file, with an ending line return character. # this file gives "23.57" as the string, including the quotes tmp = aline.split(',') appl_date.append(tmp[0].strip()) appl_price_eod.append(float(tmp[1])) print(appl_date) print(appl_price_eod) # You should be able to see the lists created from the code above. This is just our test run. #%% # Step 1 # Create a class for a stock with daily end-of-day price recorded, along with the daily volume. # class Stock: """ Stock class of a publicly traded stock on a major market """ def __init__(self, symbol, name, firstdate, lastdate, init_filepath) : """ :param symbol: stock symbol :param name: company name :param firstdate: the first date (end of list) of the price list :param lastdate: the last date (beginning of list) of the price list :param init_filepath: locate the file for date, price (eod) and volume lists """ # note that the complete list of properties/attributes below has more than items than # the numnber of arguments of the constructor. That's perfectly fine. # Some property values are to be assigned later after instantiation. self.symbol = symbol.upper() self.name = name self.firstdate = firstdate self.lastdate = lastdate # below can be started with empty lists, then read in data file and calculate the rest self.price_eod = [] # record the end-of-day prices of the stock in a list. The 0-th position is the latest end-of-day price self.volumes = [] # a list recording the daily trading volumn self.dates = [] # starts from the latest/newest date, self.delta1 = [] # daily change values, today's close price minus the previous close price. Example eod[0] - eod[1], eod[1] - eod[2], self.delta2 = [] # daily change values, the previous close minus today's close, example eod[0] - eod[1], eod[1] - eod[2] # change of the daily change values (second derivative, acceleration), # given by, for the first entry, (delta1[0] - delta[1]), # or if we want to, equals to (eod[0]-eod[1]) - (eod[1]-eod[2]) = eod[0] - 2*eod[1] + eod[2] self.import_history(init_filepath) self.compute_delta1_list() # Calculate the daily change values from stock price itself. self.compute_delta2_list() # Calculate the daily values of whether the increase or decrease of the stock price is accelerating. A.k.a. the second derivative. # END of constructor/__init__ def import_history(self, filepath): """ import stock history from csv file, with colunms date, eod_price, volume, and save them to the lists """ with open(filepath,'r') as fh: # leaving the filehandle inside the "with" clause will close it properly when done. Otherwise, remember to close it when finished for aline in fh.readlines(): # readlines creates a list of elements; each element is a line in the txt file, with an ending line return character. # ###### QUESTION 1 ###### QUESTION 1 ###### QUESTION 1 ###### QUESTION 1 ###### # Fill in the codes here to put the right info in the lists self.dates, self.price_eod, self.volumes # Should be similar to the codes in Step 0 above. # ###### END QUESTION 1 ###### END QUESTION 1 ###### END QUESTION 1 ###### END QUESTION 1 ###### # fh.close() # close the file handle when done if it was not inside the "with" clause # print('fh closed:',fh.closed) # will print out confirmation fh closed: True return self def compute_delta1_list(self): """ compute the daily change for the entire list of price_eod """ # goal: calculate the daily price change from the eod prices. # idea: # 1. duplicate the eod list # 2. shift this new list by removing the 0-th element. # 3. use the map function to find a list of delta's by subtracting the eod list from this new list. # Okay, let's try # # eod_shift1 = self.price_eod # THIS WILL NOT WORK. Try. A shallow copy. We'll talk more about that next class. eod_shift1 = self.price_eod.copy() # if you do not use the copy method here, you will get a shallow copy. # The list here is a simple list of floats, not list of lists or list of dictionaries. # So the copy() function will work. No need for other "deepcopy" variations eod_shift1.pop(0) # remove the first element (shifting the day) self.delta1 = list(map(lambda x,y: x-y, self.price_eod, eod_shift1)) print(self.name.upper(),": The latest 5 daily changes in delta1: ") for i in range(0,5): print(self.delta1[i]) # checking the first five values return self def compute_delta2_list(self): """ compute the daily change for the entire list of delta1, essentially the second derivatives for price_eod """ # essentially the same function as compute_delta1_list. With some hindsight, or when the codes are re-factored, we can properly combine them # ###### QUESTION 2 ###### QUESTION 2 ###### QUESTION 2 ###### QUESTION 2 ###### # Fill in the codes here # Need to find the daily changes of the daily change, and save it to the list self.delta2 # It is the second derivative, the acceleration (or deceleration if negative) of the stock momentum. # Essentially the same as compute_delta1_list, just on a different list # Again you might want to print out the first few values of the delta2 list to inspect # ###### END QUESTION 2 ###### END QUESTION 2 ###### END QUESTION 2 ###### END QUESTION 2 ###### return self def insert_newday(self, newdate, newprice, newvolume): """ add a new data point at the beginning of lists """ # ###### QUESTION 3 ###### QUESTION 3 ###### QUESTION 3 ###### QUESTION 3 ###### # After we have the batch of historical data to import, we # most likely will need to do some daily updates (cron jobs, for example) # going forward. There is no need to re-import the old data. # This method is then used to insert just one row of new data point daily. # We will need to insert the new date, the new eod value, the new delta1 value, # the new delta2 value, as well as the new volume data. # # insert new price data to price_eod # calculate and insert new data to delta1 # calculate and insert new data to delta2 # insert newdate to dates[] # # Fill in the codes here # # insert newdate to dates[] self.dates.insert('Something Here') # insert newvolume to volumes[] self.volumes.insert('Something Here') # insert new eod data value to price_eod self.price_eod.insert('Something Here') # calculate and insert new data to delta1 self.delta1.insert('Something Here') # calculate and insert new data to delta2 self.delta2.insert('Something Here') # # ###### END QUESTION 3 ###### END QUESTION 3 ###### END QUESTION 3 ###### END QUESTION 3 ###### return self def nday_change_percent(self,n): """ calculate the percentage change in the last n days, returning a percentage between 0 and 100, or sometimes higher. """ # ###### QUESTION 4 ###### QUESTION 4 ###### QUESTION 4 ###### QUESTION 4 ###### change = 'What should it be?' # calculate the change of price between newest price and n days ago percent = 'What should it be?' # calculate the percent change (using the price n days ago as the base) print(f"{self.symbol} : Percent change in {n} days is {percent.__round__(2)}%") # ###### END QUESTION 4 ###### END QUESTION 4 ###### END QUESTION 4 ###### END QUESTION 4 ###### return percent #%% import os # dirpath = os.getcwd() # print("current directory is : " + dirpath) # filepath = dirpath+'/AAPL_daily.csv' # lastdate is 9/12/19, firstdate is 9/12/14, # using os.path.join will take care of difference between # mac/pc/platform issues how folder paths are used, backslash/forward-slash/etc filepath = os.path.join( os.getcwd(), 'AAPL_daily.csv') # or just this should work # filepath = 'AAPL_daily.csv' aapl = Stock('AAPL','Apple Inc','9/12/14','9/12/19',filepath) # aapl is instantiated! #%% # Great! Now we can get the competitors easily filepath = 'MSFT_daily.csv' msft = Stock('MSFT','Microsoft Inc','9/12/14','9/12/19',filepath) filepath = 'GOOG_daily.csv' goog = Stock('GOOG','Alphabet Inc','9/12/14','9/12/19',filepath) #%% # ###### QUESTION 6 ###### QUESTION 6 ###### QUESTION 6 ###### QUESTION 6 ###### # # use the nday_change_percent method that you defined # Find out the stock performances in their percent changes in the last # (i) 50 days # (ii) 200 days (about 1 year) # (iii) 600 days (about 3 years) # Which one perform best in each of the periods above?? # # ###### END QUESTION 6 ###### END QUESTION 6 ###### END QUESTION 6 ###### END QUESTION 6 ###### #%% # # ###### QUESTION 7 ###### QUESTION 7 ###### QUESTION 7 ###### QUESTION 7 ###### # # Now see if the insert_newday() method works aapl.insert_newday('9/13/19',231.85,32571922) print('new dates:',aapl.dates[0:5]) print('new price_eod:',aapl.price_eod[0:5]) print('new volumes:',aapl.volumes[0:5]) print('new delta1:',aapl.delta1[0:5]) print('new delta2:',aapl.delta2[0:5]) print('last two days change: ',aapl.nday_change_percent(2)) # If the above printouts does not look right to you, make sure you fix your # insert_newday function!! # # ###### END QUESTION 7 ###### END QUESTION 7 ###### END QUESTION 7 ###### END QUESTION 7 ###### #%% # # Part 2 # This exercise to be re-done after we learn enough on Pandas. # The features in Numpy and Pandas makes a lot of things easier. # #%%