TypeError: 'value' in Matplotlib, an instance of str or bytes, not a float

1

my code:

stirs = pd.read_csv(r'C:/Users/Owner/Documents/Research/SOMA/FRED_reserves/SOMArates2.csv', \ 
na_values=" ", usecols=[0,1,2,3,4,5], parse_dates=['DATE'] ).fillna(value=0)
 stirs.info()  
stirs.columns = stirs.columns.str.strip()  
stirs['DATE'] = pd.to_datetime(stirs['DATE'],format='%m/%d/%Y')  
date = stirs.loc[:, "DATE"]  
eff  = stirs.loc[:, "EFFR"]  
repo = stirs.loc[:, "RPONTSYD"]  
rrepo = stirs.loc[:, "RRPONTSYD"]  
mrepo = stirs.loc[:, "RPTSYD"]  
mrrepo = stirs.loc[:, "RRPTSYD"]  
print(eff)  
print(mrrepo)  
x=stirs['DATE']
plt.figure(figsize=(5, 2.7), layout='constrained')
plt.plot(x, eff, label='Effective Fed Funds')  [**]
plt.plot(x, repo, label='Repos')
plt.plot(x, rrepo, label='Reverse repo')  
plt.plot(x, mrepo, label='Repos_OMO')  
plt.plot(x, mrrepo, label='Reverse Repos_OMO')  
plt.xlabel('Date')  
plt.ylabel('y label')  
plt.title("Effective FF rate, repo, reverse repo")  
plt.legend()
2

Can you indicate what fails and the full error message? And maybe make a failing self contained example? We have no idea what is in your file and if you try and reproduce the rrror you can usually tell what caused it.

3

Hours and hours trying to find this error. I’m now just guessing. I have indicated the full error. I’ve looke at the data.
Python prints out the values correctly

I did try to plot only the one variable rather than the full plot:

There are 6 columns of data: DATE EFFR RPONTSYD RRPONTSYD RPTSYD RRPTSYD
Matplotlib has no problem plotting the last 4 variable: RPONTSYD RRPONTSYD RPTSYD RRPTSYD
But the line for plotting the first variable EFFR triggers the error:
TypeError: ‘value’ must be an instance of str or bytes, not a float

EFFR is the first variable after DATE, so I’m thinking that the problem is the date format: the date in csv file is m/d/Y but python records it as Y/m/d
effrate = stirs.loc[:,“EFFR”]

I’ve deleted last row of the excel sheet, I’ve looked at the EFFR column for anomalies

I’ve tried plotting EFFR alone: and get the same error
plt.figure(figsize=(5, 2.7), layout=‘constrained’)
plt.plot(x, effrate, label=‘Fed Funds’)

When I plot all five variables, I get the same error for EFFR
fig, ax1 = plt.subplots()
color1 = ‘tab:red’
color = ‘tab:blue’
color2 = ‘tab:green’
ax1.set_xlabel(‘Date’)
ax1.set_ylabel(‘Percent (x.xx)’, color=color1)
ax1.tick_params(axis=‘y’, labelcolor=color)
lns1 = ax1.plot(stirs[‘DATE’], effrate, label=‘EFFR’,linestyle =’–’,linewidth=.5, color=‘c’)
lns2 = ax1.plot(stirs[‘DATE’], repo, label=‘Repos’,linestyle =’–’,linewidth=.5, color=‘b’)
lns3 = ax1.plot(x, mrepo, label=‘Repos_OMO’,linestyle =’-.’, linewidth=.75, color=‘g’)
ax2 = ax1.twinx()
ax2.set_ylabel(‘Level’, color=color)
ax2.tick_params(axis=‘y’, labelcolor=color)

instantiate a second axes that shares the same x-axis

lns4 = ax2.plot(x, rrepo, label=‘Reverse repo’,linestyle =’–’, linewidth=.3,color=‘k’)
lns5 = ax2.plot(x, mrrepo, label=‘Reverse Repos_OMO’,linestyle =’-.’, linewidth=.2,color=‘m’)
lns = lns1+lns2+lns3+lns4+lns5
labs = [l.get_label() for l in lns]
plt.legend(lns, labs, loc=0)
plt.title(“Effective FF rate, repo, reverse repo”)

plt.legend()

If I plot only the last 4 variables, the plot is fine

···

Nancy Hammond, Ph.D
nahammond@gmail.com
312.989.3854

4

The csv file, first few line:

···

Nancy Hammond, Ph.D
nahammond@gmail.com
312.989.3854

5

Ok, it really is the case that the TypeError: ‘value’ must be an instance of str or bytes, not a float applies to the variable EFFR.

I retyped the whole series for EFFR and still got the above error.

I need to learn how python and matplotlib treat variable types.

This is not happening to any other variable in the series.
And I’ve ruled out a problem with a date.

Time for a walk outside.

···

Nancy Hammond, Ph.D
nahammond@gmail.com
312.989.3854

6

ValueError: could not convert string to float: ‘.’

stirs = pd.read_csv(r’C:/Users/Owner/Documents/Research/SOMA/FRED_reserves/SOMArates4.csv’,
na_values=" ", usecols=[0,1,2,3,4,5],dtype={‘RPONTSYD’: float, ‘RRPONTSYD’: float, ‘RPTSYD’: float, ‘EFFR’: float}, parse_dates=[‘DATE’]).fillna(value=0)

···

Nancy Hammond, Ph.D
nahammond@gmail.com
312.989.3854

7

Last message 5/26/2022 10:36 pm

The problem converting with dtype to float is the symbol ‘.’ for missing value
Although I have converted them to NaNs

https://stackoverflow.com/questions/54065751/valueerror-could-not-convert-string-to-float-when-converting-input

stirs = pd.read_csv(r’C:/Users/Owner/Documents/Research/SOMA/FRED_reserves/SOMArates4.csv’,
na_values=" ", usecols=[0,1,2,3,4,5],dtype={‘RPONTSYD’: float, ‘RRPONTSYD’: float, ‘RPTSYD’: float, ‘EFFR’: float}, parse_dates=[‘DATE’]).fillna(value=0)

dt = np.dtype(np.int32)

DATE EFFR RPONTSYD RRPONTSYD RPTSYD RRPTSYD

stirs.replace(’.’, np.nan, inplace=True)
stirs = stirs.replace(r’^\s*$’, np.nan, regex=True)
stirs.info()
stirs.columns = stirs.columns.str.strip()

stirs[‘DATE’] = pd.to_datetime(stirs[‘DATE’])

stirs[‘DATE’] = pd.to_datetime(stirs[‘DATE’], format="%Y%m%d")
date = stirs.loc[0:, “DATE”]
effrate = stirs.loc[:,“EFFR”]

eff = float(effrate)

TypeError: cannot convert the series to <class ‘float’

I’ll work on this tomorrow 5/27/2022

···

Nancy Hammond, Ph.D
nahammond@gmail.com
312.989.3854