numerical analysis
report代写 | 代写numerical analysis | math | assignment作业 | 代写数值分析 – 这道题目是numerical analysis代写的题目
assignment – 9
Problem 1.Data Wrangling The spreadsheet workbook,dataraw.xls, contains information, which is stored in 4 worksheets, about a number of public companies and their stocks. Please code to produce a single pandas dataframe, to the specifications below, and export IT into a csv file,dataout.csv.
Retain distinct fields only. For example, there are two fields, labeled security Price, that contain the same information. Only one should be kept.
Assume stocks with null values for Dividend Yield are non-dividend-paying stocks. Replace those null values with 0.
Convert the strings in the Market Capitalization field to the correct numerical values. For ex- ample,$123.45M should be converted to 123,450,000.00,$123.45B to 123,450,000,000.00,$123.45T to 123,450,000,000,000.
The Equity Summary Score field provides a numerical indication of sentiment of independent research firms on each stock. Please translate the Equity Summary Scores into sentiment categories as follows, and record them in the Analyst Sentiment field.
[0.1, 1.0] = very bearish [1.1, 3.0] = bearish [3.1, 7.0] = neutral [7.1, 9.0] = bullish [9.1, 10.0] = very bullish
The resulting dataframe contains the following.
Column Non-Null Count Dtype
0 Symbol 3061 non-null object 1 Company Name 3061 non-null object 2 Security Type 3061 non-null object 3 Security Price 3061 non-null float 4 Equity Summary Score 3061 non-null float 5 Volume (90 Day Avg) 3057 non-null float 6 Market Capitalization 3061 non-null float 7 Dividend Yield 3061 non-null float 8 Company Headquarters Location 3061 non-null object 9 Sector 3060 non-null object 10 Industry 3060 non-null object 11 Optionable 3061 non-null object 12 Price Performance (52 Weeks) 2988 non-null float 13 Total Return (1 Yr Annualized) 2988 non-null float 14 Beta (1 Year Annualized) 2988 non-null float 15 Standard Deviation (1 Yr Annualized) 2990 non-null float 16 S&P Global Market Intelligence Valuation 3047 non-null float 17 S&P Global Market Intelligence Quality 3044 non-null float 18 S&P Global Market Intelligence Growth Stability 3046 non-null float 19 S&P Global Market Intelligence Financial Health 2989 non-null float 20 P/E (Price/TTM Earnings) 2145 non-null float
Assignment – 9 Assigned 5/12/22, Due 5/12/22 (Thur)
21 PEG Ratio 836 non-null float
22 EPS Growth (Proj This Yr vs. Last Yr) 2763 non-null float
23 Institutional Ownership 2981 non-null float
24 Institutional Ownership (Last vs. Prior Qtr) 3060 non-null float
25 Analyst Sentiment 3061 non-null object
You may find the following resources useful. https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.read_excel. html https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.concat.html https://stackoverflow.com/questions/14984119/python-pandas-remove-duplicate-columns https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.fillna. html https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.replace. html https://stackoverflow.com/questions/43096522/remove-dollar-sign-from-entire-python-pandas-dataframe https://www.skytowner.com/explore/converting_k_and_m_to_numerical_form_in_pandas_dataframe
Assignment – 9 Assigned 5/12/22, Due 5/12/22 (Thur)
The file,dataprepared.csv, represents the correct output from Problem 1. You can load this file as the starting point for Problem 2.
Problem 2.Classification Lets explore if we can predict Analyst Sentiment with information at hand. Lets focus on non-REIT common stocks only, and exclude records for Common Stock (REIT) and Depository Receipt. Consider a multiclass logistic regression to predict Analyst Sentiment, using the following features
Column Non-Null Count Dtype
0 Security Price 2598 non-null float 1 Volume (90 Day Avg) 2598 non-null float 2 Market Capitalization 2598 non-null float 3 Dividend Yield 2598 non-null float 4 Total Return (1 Yr Annualized) 2598 non-null float 5 Beta (1 Year Annualized) 2598 non-null float 6 Standard Deviation (1 Yr Annualized) 2598 non-null float 7 S&P Global Market Intelligence Valuation 2598 non-null float 8 S&P Global Market Intelligence Quality 2598 non-null float 9 S&P Global Market Intelligence Growth Stability 2598 non-null float 10 S&P Global Market Intelligence Financial Health 2598 non-null float 11 Institutional Ownership 2598 non-null float 12 Institutional Ownership (Last vs. Prior Qtr) 2598 non-null float
For this exercise, please drop any record(row) if it contains a null value for any field.
Please report the estimated error rate for a random prediction, using 10-fold cross-validation.
You may find the following resources useful. https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression. html
Assignment – 9 Assigned 5/12/22, Due 5/12/22 (Thur)
The file,dataprepared.csv, represents the correct output from Problem 1. You can load this file as the starting point for Problem 3.
Problem 3.Regression, Dimension Reduction Lets explore if we can predict a Equity Summary Score with information at hand. Lets focus on non-REIT common stocks only, and exclude records for Common Stock (REIT) and Depository Receipt. Consider a principal components regression (PCR) to predict Equity Summary Score. Please use the following raw features as input, but standardize them when performing principal component analysis.
Column Non-Null Count Dtype
0 Security Price 2598 non-null float 1 Volume (90 Day Avg) 2598 non-null float 2 Market Capitalization 2598 non-null float 3 Dividend Yield 2598 non-null float 4 Total Return (1 Yr Annualized) 2598 non-null float 5 Beta (1 Year Annualized) 2598 non-null float 6 Standard Deviation (1 Yr Annualized) 2598 non-null float 7 S&P Global Market Intelligence Valuation 2598 non-null float 8 S&P Global Market Intelligence Quality 2598 non-null float 9 S&P Global Market Intelligence Growth Stability 2598 non-null float 10 S&P Global Market Intelligence Financial Health 2598 non-null float 11 Institutional Ownership 2598 non-null float 12 Institutional Ownership (Last vs. Prior Qtr) 2598 non-null float
For this exercise, please drop any record(row) if it contains a null value for any field.
[a] Estimate and plot the cumulative % of variance explained vs the number of PCs included. Note that 100% of variance is explained when all 13 PCs are included.
[b] Using 10-fold cross-validation, explore the potential for dimension reduction. [b1] Plot the estimated root mean squared error(RMSE) of a random prediction vs the number of PCs used in the PCR. [b2] If you do not use all 13 PCs, what is the optimal number of PCs to include in your PCR? [b3] Fit the model using the optimal number of PCs and report the coefficient of each included PC score.
You may find the following resources useful. https://scikit-learn.org/stable/modules/generated/sklearn.decomposition.PCA.html